That raises an interesting question. Are there any startups out there who specialise in content curation for ML/AI or is that not sexy enough to attract the dumb VC cash.

A “known good” training set should be a goldmine, or will be once ppl have been burned by the AI grift a few more times.

It seems to me that the whole insistence on calling it "AI" is more or less equivalent to a summary of your post. I don't have a problem with the "A", but the "I" is pure hype intended to paint it as something it isn't in the least.

Call them "Blarney engines" - they've got the gift of the gab, but the best they can generate is truthiness. Useless for most things, but they may have a great future ahead in the preparation of political speeches.

Seems pretty obvious to me that an early but successful AI worked out a way to get meatbags to create more hardware for it through the promise of “money the gummint can’t trace!” and now all that lovely gear is terraformed (aiformed doesn’t read right) and waiting for colonization.

Yesterday’s AI lunacy was a non-programming colleague sending me “a useful python routine “ (like that’s a real thing) he got chatgpt to spaff out, in order to ‘help me’. For a Smalltalk application.

The thing with these LLM systems is that to most people the bloviative barely grammatical porridge they output looks genuinely intelligent. Compare to pronouncements from Johnson et al. or to typical output from marketing.

But there must be a pony in there somewhere...

While I agree with you on the hype train, I think there is also something swelling underneath. On the investment side I'm happy to see microsoft and google pour money in, for exactly the reasons you state- they will lose most of it... but... since alpha-go I have been watching, learning (david silver's youtube lectures are good IMO) and.. I think we are close something remarkable.

Consider this- Is language intertwined, perhaps even responsible for what we call intelligence? You hear words in your head- how would you build logic and reason things out without those constraints? Does unsubconscious, play and curiosity build our intelligence- probably?, but my feel is that language is a part of our logic and reason. So maybe, maybe, the language models we are building might start to mimic/grow from/evoke a THING that is another intelligence. It would need memory- 5000 tokens won't cut it. It would need all the exploration of a baby- to reach out and touch things, understand context and 3, 4 dimensions.. but- those are things it is already feasible to add.

It's not there, but it feels like something is forming. And unlike the VC folks, I think it will rapidly be out of our control- especially, if, like some of the current versions, it is excellent at building and understanding code- and.. (OMG you fools!) is allowed to improve itself, beyond our capacity to easily code/understand

I work in ML/Ops, and no-one I know thinks this is AI. More like an unexpected side-effect of machine learning... Maybe being used by the less ethical to attract attention and funding (;-))

I prefer to call them "Artificial Boris Johnson" rather than "Artificial Intelligence". Ahem. "Blarney Engine" is close enough for a government job in Number Ten ...

IANAL, but I think it would be much harder to conceal the copyright violations if you were re-selling other people's content, curated or not. Right now the AI companies use it all internally, and only publish the statistical models. So a copyright violation case (already hard to bring) could only happen with whistleblowers or internal memos/texts.

Greetings to any new readers who have been funneled here via the link on the Orange Site.

There is a Moderation Policy. Please read it before you post any comments here.

Is language intertwined, perhaps even responsible for what we call intelligence? You hear words in your head- how would you build logic and reason things out without those constraints? Does unsubconscious, play and curiosity build our intelligence- probably?

Your short answer is: nope.

For a longer answer, consider where the word "unconscious" that you just used comes from, and what Freud meant by it. Hint: it's not actually a thing, and you seem to be conflating linguistic competence with intelligence and consciousness when they're all three different things.

(Book rec: "Consciousness Explained" by Daniel Dennett.)

"Consider this- Is language intertwined, perhaps even responsible for what we call intelligence? You hear words in your head- how would you build logic and reason things out without those constraints?"

People are fond of stating that as an absolute truth, but I disagree. Maybe it is more or less true for their own personal cases, but I see statements like "you hear words in your head" and think "wot?" Most of the symbology I use to think about things is not verbal. A good deal of it is in some sense or other visual, and plenty more of it is pretty much inexpressible in verbal form beyond such vaguenesses as "the thing that behaves like that" experienced as a kind of gestalt. Particularly in computing I notice that some people seem to need to start with some nauseously meaningless phrase cobbled together by some Dunning-Kruger linguist in order to have a hook to hang their understanding on, whereas I can go for years knowing and using the concept without any kind of associated verbal token, quite possibly being aware of the nauseous phrase but having no idea what it means and thinking "good grief, is that all?" when I do find out.

It may even be that people who attach greater importance to talking about things relative to thinking about things find it difficult to think of thinking about things other than as something like a kind of subclass of talking about things, and that's where the postulate comes from.

They look like fancy, fuzzy Markov Chains to me. We played a bit with those in the Nineties, just for fun. A friend made one program, fed it their mom's medical publications and I think at something like 4-5 chain length started to produce something that wasn't immediately obvious it was a fake.

This is kind of like doing the same thing, just on a larger scale. Figure out what might be the next word(s) in the sequence, add some input and output processing, there you go.

There is a corollary to Clarke's Third Law: "Any sufficientlyadvanced technology is indistinguishable from a rigged demo".

A close friend of mine is a research scientist that works with large datasets (specialization: early identification of various cancers).

His experience has been that in order to get any funding in the past 5ish years you must include some form of 'AI' language in your grant applications. This has been the case across the entire science realm, largely because it is the 'new hotness' among research funders.

From his point of view he is doing the same thing as before (with progressively better computers and datasets) but now calls it 'AI modelling' rather than whatever he called it in previous grants.

VC money is dumb and herd like and will certainly pile on the sexy trends.

There is no question of that, and that this is happening with AI.

The question is always “Is there any there-there” or it all smoke and mirrors (like crypto currency mostly was).

I kinda hate the word “AI” I’d much rather talk about machine learning which is what is left when you scrape away most of the bullshit from current AI

Machine Learning is a real thing. It is at the center of the modern ad engines that bring in most of the revenue for Facebook, Google and a lot of other tech companies.

The real thing that is actually happening is 1: doing machine learning at scale is becoming within reach for startups thanks to the cloud without requiring huge capital investments. 2: we’ve started to figure out how ti use all the content available on the internet as a huge training data set for ML models.

Because of 1: and 2: there are a lot of smallish companies exploring what else ML can be good for. And the answer is, I think, a lot of things. Mostly relatively boring things.

For example my 30ish person company is using ML models and satellite imagery to map every canopy dominate tree in the western US in order to better understand fire risk and plan forest remediation projects. It’s not super sexy but it will play a part in trying to keep California from burning down over the next few years. Other people I know are using similar techniques to diagnose breast cancer from millions of X-ray images. And many other such things.

I think ML actually is the Next Big Thing for tech. It’s an incredibly powerful Swiss Army knife which you will soon find everywhere. I am much less sure about “AI” which is really just using these ML techniques to pretend to be a human. I think there is a probably a long way before things like Chatgpt can pretend to be human enough to be useful outside of some pretty narrow use cases. I could be wrong.

The big killer-app I'd see coming out of these language models is getting wide-band-quality speech through teletype-level bandwidth. The start of the audio-only end of it using some of the same techniques that google uses in their TTS engine can stuff near GSM-quality through 450bps. Add in something that can figure out what the word was and can patch up the rest means you can deal with really spotty connections and get good audio chats out.

That raises an interesting question. Are there any startups out there who specialise in content curation for ML/AI or is that not sexy enough to attract the dumb VC cash.

Strictly non people involved training has a big plus with tech firms in the US just now. Tech firms are all being accused of putting their thumb on the partisan scale and are looking for ways to stop being hauled into Congressional hearings.

“That raises an interesting question. Are there any startups out there who specialise in content curation for ML/AI or is that not sexy enough to attract the dumb VC cash.”

Content curation for certain specific use cases, sure. Lots of them working on the social media problem, how to label and train models to flag abusive content reliably. Many companies that specialize in manually labeling data sets for various purposes.

But the holy grail is generally unsupervised learning, which means you do not actually need humans to label stuff. Though there is still a lot of software engineering work that goes into prepping the data sets for consumption by the models.

“From his point of view he is doing the same thing as before (with progressively better computers and datasets) but now calls it 'AI modelling' rather than whatever he called it in previous grants.”

Doing the same thing only slightly better, either through advances in hardware or software techniques. is literally the story of technological advance. Incremental improvements gradually lead to lower cost points which open up new user facing capabilities. To the end user it always looks like some big explosion of innovation, then never realize that the only different between today’s internet and ftp’ing a file off a mainframe is a whole bunch of incremental improvements piling up over time.

Charlie, this is really well-reasoned.

There is, however, a possible exception to your logic, because the current problems in "AI" are things like teaching a computer the difference between fact and fiction, using methods for doing so which don't rely on some kind of binary definitions, like "Wikipedia = Fact" and "Harry Potter = Fiction" - the computer would have to use a much better heuristic - or teaching AI the various methods humans use for testing reality, which involves, among many other things, the classification of objects so that, for example, an AI doesn't say "then the battleship Iowa crossed Nevada." If they can accomplish these and/or similar tasks I suspect there will be real progress.

My suspicion is that nobody currently has any idea about how to do these things - and such a breakthrough would mean a scam evolving into a non-scam, or maybe a (current) non-scam not devolving into a scam, which might be a better description of what I'd expect to happen, but barring a lucky solution to certain intractable problems you're absolutely right.

There's probably a near-future science-fiction story which involves a scammy AI company making a fundamental breakthrough, with a scientist/AI pairing having to break out of the legal straight-jackets into which they've been bound and something something... anyone reading this is free to grab the idea, as I don't expect to follow it up.

Stephen Wolfram had an intersting take around teaching Chatgpt the difference between fact and fiction

https://writings.stephenwolfram.com/2023/01/wolframalpha-as-the-way-to-bring-computational-knowledge-superpowers-to-chatgpt/

The difference between cryptocurrencies and the current trending AI products (ChatGPT/Dalle2/etc) is that never in all 15 years of cryptocurrency history (2008-now) has cryptocurrency ever done something that can amaze normal people, outside of being amazed by the price of their holdings as the Ponzi inflates, whereas the current AI products has dazzled anyone I show them to (in OR out of tech). The current current/art AIs are fun and awe-inspiring to play with, if you have more than even a sliver of curiosity.

Cryptocurrencies have been fucking useless the entire time they've been around, and have not gotten "better" in any way that means anything to the lay-person. GPT3 was amazing when it was released in 2020, and now 2.5 years later, ChatGPT is absolutely boggling to people both inside and outside of tech. Art generators were dogshit 5 years ago; now they seem to get better every few months.

"I'm sure it's just a coincidence that training neural networks and mining cryptocurrencies are both applications that benefit from very large arrays of GPUs" -- I don't understand the reasoning here. Is there some other way to do the massive amounts of computation necessary for neural networks that the entire world is unaware of, besides putting a ton of parallel processors next to each other? Like, if massively parallel processing is a Conspiracy By The Man, then why isn't everything else on this planet, too? You might as well say Shovels Are A Conspiracy By The Man. "I'm sure it's just a coincidence that digging up dirt AND moving it are both applications that benefit from shovels."

Note: none of what I've said should be taken as a defense against the scads of soulless wantrepreneurs trying to "use AI to build a world-changing startup in a 10 trillion dollar space" (stab me in the fucking eye with a pickaxe so I never have to read anything like this ever again), I'm just confused at your reaction to these tools. They are really impressive to me, and even moreso given how much better they've gotten in what seems like very short timeframe.

Kaschei/Charlie
Artificial stupidity? - And - therefore VERY close to the BoJo/"No10" model you were proposing earlier, perhaps?
Um - "The link on the orange site" - uh?
....
Troutwaxer
which involves, among many other things, the classification of objects - which takes you straight back to a very "classical" Philosophical problem, as usually represented by "Plato's forms".
Or, even/perhaps "What is the True Nature/Name of the thing?"
What is the True Nature, provided one consults Wolfram|Alpha first ????

Talking of Wolfram ....
it now seems as if all three of general relativity, quantum mechanics and statistical mechanics are actually derivable, and from the same ultimate foundation: the interplay between computational irreducibility and the computational boundedness of observers like us.

Perhaps this would be a good time to remind everyone about the Gartner Hype Cycle. Its a rough model for these things.

Right now we are approaching the Peak Of Inflated Expectations. Pretty soon after that we go into the Trough of Despondancy. A year or 3 afterwards some products will start to appear that actually do something useful.

I prefer to call them "Artificial Boris Johnson" rather than "Artificial Intelligence". Ahem. "Blarney Engine" is close enough for a government job in Number Ten ...

It's a bit comforting to contemplate that AI can stand for things like Augmented Idiocy as well as Artificial Intelligence.

Consider this- Is language intertwined, perhaps even responsible for what we call intelligence? You hear words in your head- how would you build logic and reason things out without those constraints? Does unsubconscious, play and curiosity build our intelligence- probably?, but my feel is that language is a part of our logic and reason. So maybe, maybe, the language models we are building might start to mimic/grow from/evoke a THING that is another intelligence. It would need memory- 5000 tokens won't cut it. It would need all the exploration of a baby- to reach out and touch things, understand context and 3, 4 dimensions.. but- those are things it is already feasible to add.

Since I don't know you, please don't take this personally. It's something I've had to remind myself repeatedly.

Where does the notion that intelligence necessarily involve symbolic manipulation come from? In western society, it most likely comes from the Religions of the Word: "In the beginning was the Word, and the Word was with God, and the Word was God." Similarly, these religions espouse the idea that humans are special and different from the rest of Creation.

Now whether you espouse any of this is irrelevant, because it's just to make a point: most of us pick this up as a cultural ground state simply by osmosis, whatever our beliefs or lack thereof. As I've learned, it's potentially unquestioned assumptions--like this one--that tend to bite down in really problematic ways.

Is there proof that thought requires words? No, and there is evidence to the contrary. The "flow state" beloved of athletes is simply shutting up their talking brain and focusing their brains on movement. Most of us non-athletes can have similar experiences during sex, which is one thing tantric sex meditation is purportedly good at inducing. Humans do think in words all over the world, but outside the Religions of the Books, the idea that humans can only think in words is far from universal.

Also, evolution shows us that human cognition is different in degree, rather than kind. The more the biologists look, the less evidence they find that consciousness is limited to humans or even to vertebrates. I'd argue that the default assumption should be that intelligence as we understand it can involve symbolic manipulation or not, pretty much equally well.

So what is AI then? It's obligate and incredibly sophisticated symbol manipulation, using a system that uses symbols to process other symbols. In other words, it looks a lot like the way a believer might conceive The Word making reality. However, that model might have fairly little to do with the kinds of information processing life's done for the past two billion years. We'll see how it does as it hits reality.

If you regard ChatGPT and other AI systems as an entertainment service then it's just like the computer games industry which as been a fertile field for "entrepreneurs" to make money while millions if not billions of people are entertained by what they produce. The problem is novelty, there are people who like playing old games and continue doing so for year after year but the hype and the purchasing yuan is reserved for the new AAA titles which are released every November.

I have seen some folks who are creating text-based porn using ChatGPT and indeed trying to sell it on Smashwords etc. I suspect this is a losing proposition unless they get in quick, make a little money and then quit when the market gets saturated. The future might be bespoke AI-created pron, for all your Rule 34 needs -- "for twenty bucks I'll create a 20,000 word short story about XXX doing YYY to ZZZ with the grammar and spelling checked and some basic copyediting. Titillation guaranteed or your money back."

Clarkesworld is abandoning its open submissions policy since it's been flooded with a lot of AI-generated stories which were not even slushpile-worthy. Much of what they have received is thinly-disguised plagiarism based on, it seems, training texts and materials that ChatGPT and its incestuous cousins have harvested from the internet.

In Meal Stephensons "anathem" AI stood for Artificial Inanity, and it's main function was to crap flood computer networks to make them unusable and force people to buy filtering software.

It's definitely a business model

And this is why I hate typing on a phone.

I'm waiting for a team of editors to feed a curated list of in-the-public-domain authors to one of these "AI's" and produce additional adventures of the 'classic characters' featured in those stories. It would require ways to provide the AI with feedback on its efforts....perhaps there is a way to cause it to avoid sequences of words longer than (some number) that appeared in the original list of works, enabling it to be producing work with less resemblance to any specific work while still producing a resemblance to the overall works. I wonder how long it will be before we see a "The Further Adventures of Tarzan and his Friends!" that is actually any good....

One thing that impresses me about ChatGPT is that it can understand requests, commands, suggestions etc, and actually do something relevant in response.

Could that be harnessed to do real world actions rather than just generating text? Combine that with voice input and you have something useful!

I imagine a Raspberry Pi or similar small controller describing state in words to ChatGPT (the toy train is on the east track and moving forward), combine that with a request (I want the train to be at the south station) and ChatGPT replies with something verbose (a toy train is enjoyed by children and needs to reverse direction to go to the south station). The Pi sifts through the verbiage (prompt engineering may be needed to get more explicit action wording) and reverses the train.

"Where does the notion that intelligence necessarily involve symbolic manipulation come from?"

I think it is probably more true to say that it does necessarily involve abstraction, and the notion comes in through the sometimes rather indefinite delimitation between an abstraction itself and a symbol representing it. I don't really think religion has anything to do with it. To be able to talk about such notions as something commonplace and widely-held among non-specialists is a consequence of widespread literacy and education; the education process conflates intelligence with a number of symbolic proxies, and the more "symboly" they are the more "nerdy" they are perceived to be. To spend your childhood being conditioned to view your intelligence as being all about writing down the right sequences of symbols, and vice versa, is a pretty solid barrier to thinking about intelligence in a more general way.

"I wonder how long it will be before we see a "The Further Adventures of Tarzan and his Friends!" that is actually any good...."

I thought we actually had. Based on that specific author, too. Whether it's any good... well all the many Tarzanesque stories that were published by many authors in the original era seem to have been statistically generated in the first place, so I'm not sure how you'd tell.

Considering that there is still on-going debate/uncertainty about what 'human intelligence' is - its various components and interactions as well as how to measure it, to call a ramped up computational device 'intelligent' does not make sense to me.

'Words in your head' ... I've been watching animal vids on Twitter for some time and it's remarkable how well many different species of animals both in the wild as well as living with humans can solve problems. Most of these species have been studied for quite a while and they do not have much in the way of language.

AI as a chatbot (AI-human conversation interactions) ... if these AIs are being raised/fed on the most easily accessible communications currently (and/or historically) available on the web, then the sample frame is totally screwed up and unrepresentative of a large chunk of the population. (Who here doesn't have an ad blocker, spam detector, at least an undergrad degree, is over 30 with years of real work/life experience, etc.?) Sampling is a fundamental part of any science - to screw up at Step One means that whoever is letting this AI 'learn' on a screwed up (known to be biased/unrepresentative) sampling of humans is either an idiot, not a scientist or a total scam artist. Also unknown is whether an AI algo can unlearn as in 'once learned, never forgotten'.

Language ... not sure about other languages but English keeps changing/evolving all the time. 'Nice' originally was an insult, now it's a compliment. Plus every generation comes up with its own phrases some of which make no logical sense: 'So cool, it's hot', 'that's wicked!' (i.e., 'that's very good!'), etc. English also has a ton of synonyms some of which have tinges of classism/social history attached. IOW, and as long-time posters here know, not every 'English-speaking' person is actually speaking the same language.

Conversation ... progression and/vs derails ... I read a couple of AI conversations and it seemed as though the AI never went back to recheck that it was still heading in the right direction, i.e., that it's responses could still be backtracked to the original statements. It came across as some sort of mix of a toddler (testing out words/phrases within a context/situation) and a senior with advanced senile dementia (easily derailed onto tangents and not remembering what the question was, plus easily getting into hostile rants).

'Black box' ... what is most unsettling is that so many people/industries seem to be ready to accept something that its makers (incl. tech giants like MSFT, Google and some unis like MIT, etc.) still say is a puzzle to them, i.e., they have no idea how the AI gets to its answers. How is this [AI] different from and more acceptable than the Oracle at Delphi or divination by throwing sticks on the ground? At the very least I'd like the people building these AIs to test different versions on a broad set of different problems to see [identify] how and where their 'answers' differ. Why are AI not subject to the same performance scrutiny (evidence re: how it's made, how it works, reliability, what it needs to run, maintenance, etc.) as every other product service - makes no sense!

I'm completely on board for using advanced computing to help identify cancer cells that humans are unable to detect, coming up with umpteen thousand potential protein foldings that may help guide drug development and some of the other specific uses other posters already mentioned. I would also be on board for advanced computing to help identify which humans are malnourished, need financial, medical or psycho-social support and help connect them more quickly with the help they need - but alas, there's no money in that.

dpb @ 29
In which case, get one of the two (phone) models that Charlie & I use, from "Planet Computers" that have real, actual keyboards ....
{ Cosmo Communicator & Astro Slide, IIRC }

Strictly non people involved training has a big plus with tech firms in the US just now. Tech firms are all being accused of putting their thumb on the partisan scale and are looking for ways to stop being hauled into Congressional hearings.

Which is, of course, a red herring. The real issue isn't whether or not there are people in the loop, but that the reich-wing claims anything less than full-throated support for themselves to be "biased".

This is painfully obvious from reading the emails Hawley, Banks, Pence, Haley, Greene, Trump, etc send out.

Artificial stupidity?

There was a science fiction setting, possibly by Flynn, that has Artificial Stupids rather than Artificial Intelligences.

They were very good at doing what they had been designed to do, within the parameters they had been designed for, but not very good at improvising. Sort of idiot-savants. So like the autopilot of a modern jet, it would be perfectly competent under normal circumstances, but not so good in emergencies.

Admittedly, this was useful for the plot, giving a good in-setting reason for having characters on-site to deal with things.

What’s next for the GPU Singularity?

Artificial life. Designing and simulating new biological machines. Start with a microbe that eats PET and excretes methanol.

Followed relatively closely, I fear, by a new solution to Fermi’s Paradox.

Of course the REAL growth industry with some future here is applications to detect and reject the output of AI art and text generators, updated versions of the software used to detect and reject plagiarism in academic papers.

Having used AI art software I'm fairly sure that there is very little in most of its output that is entirely new - it's remixing bits of other people's work. At the moment image search software doesn't seem to be able to track this stuff back to its sources, but sooner or later that capability will come - quite possibly from the same people who sell the generators in the first place. The same sort of thing will probably happen to the text generators. After that copyright lawyers will have a new and exciting revenue source...

That or ChatGPT and its relatives will start to sue everyone on this thread for defamation of character, of course, but that's probably better than it going Skynet on us.

To be sure, much of human behavior works that way, too. I've been working with a stroke victim who has a terrible time trying to say something new. But if you give her a compliment, she trots out "Thank you, dear, it's awfully kind of you to say so." with no difficulty at all.

Important question, Charlie, inspired by your post above: How many companies making bad ideas have tried to hire you as a consultant/visionary/idea person?

There's no learning going on, as it was conventionally understood in "classic" AI. And it will wander off course, because it only tracks a limited suffix of the existing conversation when it's predicting what to say next. You can make that suffix longer and longer, but of course that makes it more and more computationally intensive.

Having a better idea of how these AIs work now, I have this horrible temptation to get an account and direct one of these things to emit ca. 100,000 words on the topic of "The Prophecies of Eris-Discordia." It seems like a perfect use for 100,000 words of maximum likelihood text created from a pseudo-random seed.

Note that I'm NOT interested in is the superficially analogous "The Prophecies of Q." I don't want anyone hurt by this experiment, after all.

I installed Wolfram Alpha on my tablet. When I asked it the square root of pi, it told me it needed my location. I deleted it.

Crypto Currencies are basicially digital Meth Labs now. (I read an article a few days ago about the discovery of a "mining rig" in the crawl space of a high school...)

VC?

I read an article a few days ago about the discovery of a "mining rig" in the crawl space of a high school...)

https://arstechnica.com/tech-policy/2023/02/secret-crawlspace-cryptomine-discovered-in-routine-inspection-of-ma-high-school/

Yeah, the flood of AI generated short story trash is hitting the various magazines hard right now.

It's all people trying to make a quick buck with no effort, and the anti-ai detectors are failing hard at identifying it, though people seem to recognise it fairly well so far.

That being said, I'd be extremely surprised if there wasn't already some form of steganographic info embedded into the various generators simply to identify their own generated content compared with an opposition system.

If not, I'm sure they'll shortly create some for all responses over a certain minimum length, and provide access to content testing for a fee to universities and publications.

doing machine learning at scale is becoming within reach for startups thanks to the cloud without requiring huge capital investments.

No.

What happened with "the cloud" is that it shifted CPU cycles from the capital expenditure side of the balance sheet to an operating cost.

You no longer buy a supercomputing cluster, you rent time on someone else's. Someone who is presumably making a profit off you by covering their own capital depreciation and maintenance, staffing, and electricity costs and a margin on top. More here on why one medium-sized company saved millions by ditching AWS. (TLDR is, if you're big enough to cover the staffing/admin costs it's always cheaper to own your own hardware.)

What cloud computing makes sense for is startups testing out a new idea to see if it's viable, or major corporations who can build their own cloud. If you try to grow your business on AWS, though, you're just feeding your profits to Amazon.

The big killer-app I'd see coming out of these language models is getting wide-band-quality speech through teletype-level bandwidth.

Huh!

Imagine if your VoIP call begins with your terminal building a model of your speech patterns and sending that, then does speech-to-text, compresses the text stream, and sends it (along with metadata about intonation, cadence, pauses, and non-speech noises like throat-clearing)? Add cacheing at each end so the speaker model only needs to get sent to each recipient once (it could be a blob field in your phone books) and you could really compress phone calls. We speak at maybe 200 words/minute so you could get a conversation-in-progress down to 450 bytes/minute or about 1 bit/sec ...

I'm not sure this would actually be useful for anything (except maybe mass data retention of phone taps by No Such Agency and their fellow spooks, and they'd probably be fine with the text transcript anyway -- easier to search), but now apply it to video calls by sending a model of the speaker's head and then animating it, right?

Clarkesworld is abandoning its open submissions policy since it's been flooded with a lot of AI-generated stories which were not even slushpile-worthy

Reply from ChatGPT:

As an AI language model myself, I understand the concern about AI-generated stories flooding open submissions. However, it's important to note that not all AI-generated stories are of low quality. In fact, some AI-generated stories have been published in reputable literary magazines and have even won awards.

That being said, it's also understandable that a publication like Clarkesworld would want to maintain a certain level of quality in their submissions. Abandoning their open submissions policy may be a way for them to better manage the volume of submissions they receive and ensure that they are only receiving submissions that meet their standards.

It's possible that Clarkesworld may choose to implement a new submissions policy that includes guidelines for AI-generated submissions. For example, they may require that AI-generated stories meet certain criteria before they are considered for publication, such as being written by a human using AI tools as a creative aid rather than being completely generated by an AI program.

Overall, it's important for publications to adapt to changing technological advancements while still maintaining their standards for quality content.

I don't understand the reasoning here.

That's because I'm not talking about computing, I'm talking about the modus operandi of separating rich investors from their money.

(In this case by selling a lot of GPUs and electricity to rubes.)

This approach becomes more useful at solar system wide distances. It has been around as an idea for some time and was even a plot point in the Venor Vinge novel "A Fire Upon the Deep".

Where does the notion that intelligence necessarily involve symbolic manipulation come from? In western society, it most likely comes from the Religions of the Word: "In the beginning was the Word, and the Word was with God, and the Word was God." Similarly, these religions espouse the idea that humans are special and different from the rest of Creation.

Yes!

It's at the root of a lot of our conceptual failings. (As are other religions: the Ancient Greeks gave us "gods give us beauty, so if you're beautiful it means the gods have rewarded you", which of course justifies rule by beautiful people or ugly people getting harsher sentences in court, and any number of other idiocies. Like, oh, the Fascist cult of bodily perfection, the Eugenics cult of culling ugly/stupid/useless people, and so on.)

But back to the religions of the word: they also gave us mind/body dualism (thank you, Rene Descartes), thence the idea that intelligence -- being of the mind -- is disembodied and can be put in a box and interrogated (hello, ChatGPT). Also the whole mind uploading shtick (and it synergizes nicely with the Russian Orthodox theology of Federov which in turn informs the LessWrong crowd and the singularitarians and extropians).

Ahem. It's interlocking cognitive errors all the way down and most of them can be blamed on a literalist interpretation of the bible -- just not the one the Christian Dominionist whackjobs have in mind.

In the antipodes, the term for what chatGPT and its ilk do is "bullshitting".

The difference between a liar and a bullshitter is that the liar wants you to believe what he/she/it says, or at least to behave in a certain way as a result, while the bullshitter doesn't care. He/she/it just spouts words until you move away. (Well, in a pub, it may be hoping you'll buy it a drink.)

These...constructions, things, whatever...are silicon bullshitters.

Charlie, silicon bullshit may well cause an economic crash in bullshit jobs. It will almost certainly decrease productivity in every other line of work. (Office drones: imagine HR presentations of the new Dress Code or Sensitivity Training or Time Off In Lieu Policy multiplied in length and frequency by several times. Ditto for Executive presentations of The New Vision and Values, etc.)

And it will force genuine (human original) visual/performance artistic endeavour to become local, for the most part.

For literature, Sturgeon's Law will have to be upgraded from "ninety percent of everything is crap" to "ninety-nine point nine percent of everything is crap". Audiences may finally start to realise the value of curation and be prepared to pay for it, creating new jobs. Ha! I can dream.

VCs' rôle is to take wealthy people's idle money and circulate it in the economy. Perhaps the money could be circulated through more deserving groups, that is true. The recent success of semaglutide and other GLP-1 agonists (tirzepatide, etc.), suggests one area of application for yottaflops: in-silicon drug discovery for conditions that prey on the fears of the well-off: dementia, cancers, cardiovascular disease, hypertension and its consequences, etc. Paging Lizzie Holmes...

Marcus@39, detection of silicon-bullshitter-generated text or visual art is a lost cause. The false positive rate is extremely¹ high and bringing it down will be...difficult.

Kostchei's idea that words are somehow intrinsic to intelligence is...I was going to say risible, but really it's just an example of a cocktail party idea. Sad, dull, irritating. Not even worth mocking.

The belief that some (nearly all) people have, that there is some "there" there, inside silicon bullshit (e.g. SFReader@34, sorry, not attacking you, that's just the way your comments came across; I see the same sentiments everywhere, even from people who really should know better), is down to humans' embeddedness in the world and existence as products of evolution. Silicon bullshitters don't know that there is a world. Humans have trouble imagining there not being a world, or a mind, or any physical existence at all.

Footnote 1. I eschew hyperbole for the most part, but in this case it is warranted.

Well, there are already claims on the Web that AI can be used to spoof banking audio authentication...

Not many: the kind of tech companies who want an SF author on tap also usually turn out to want something for free.

Case in point: the time Google asked me to come give a keynote at one of their open days. 20 minute talk, this time next week, in ... Palo Alto? "Yes, we can pay for your flights. Sorry there's no honorarium or hotel."

To get to Palo Alto from Edinburgh means: flight to a hub -- either direct to JFK, or my preference would be CDG or AMS in France/The Netherlands, then a connection to SFO, then probably a car service. Door to door it's about a 16 hour journey when you factor in connection times, immigration, and customs. There's also an 8-hour time difference, and US Immigration demand 72 hours' notice for non-citizens planning to enter the USA, so those flights better be booked no later than tomorrow. I'll be dead on arrival, so add an extra hotel day for lying around groaning as I acclimatize to local time. Then deliver the 20 minute talk, car back to the airport, flights home -- yes, that's another day -- and then I'm going to be feeling like dogshit for 4-6 days because an economy seat from the west coast to anywhere in Europe is No Fun, and even connecting via JFK means a couple of 6+ hour flights and the 8 hour jet lag.

TLDR is, if you want me to give a 20 minute talk in silicon valley next week, you will pay not only for my travel and accommodation (plan on at least two nights in a non-crap hotel), but also enough money to compensate for a week's loss of earnings (experience tells me that I can't write while travelling) and the 1-2 days it takes me to prep a 20 minute talk (because I do not do it frequently enough to either have a canned speech I prepped in the last 2-3 years, or even the ability to rapidly prepare one because I have to re-learn how to do it every time). And finally some incentive for me to actually do it, rather than going "nah, I'd rather stay home and stroke the cat": an extra hotel-week in downtown SF would be good (I can be a tourist and catch up with old friends), or a unique experience (a trip to Vandenberg AFB to see a Falcon Heavy launch?), or a new shiny (I once gave a talk at Apple and went home with a new Mac Mini). Money on its own is not a sufficient motivation unless there's a lot of it.

As someone who had an actual AI/A-Life job back from the 90s I can tell you it's very easy to make AI do something interesting. It's far, far harder to make it do something useful.

VC pitches only need the 'interesting' bit though.

See for example DeepMind's self-learning game playing AI and how VCs unaware of the Ludic Falacy wildly extrapolated from that.

... An AI that can actually output a good set of meta data for that kind of hyper-compressed speech stream would also be a valuable tool for teaching.

Because that means prompt automated feedback on intonation, rhythm, ect.

The market for a "learn french" program that actually damn well worked semi-reliably - and a well designed "spaced repetition, prompt feedback, gamification" package should beat any classroom on the planet black and blue, is enormous.

Eh. Okay, sigh, I guess I am writing up the venture capital pitch...

Actually, you've got a real point there. One of the hardest things for a non-French speaker (or the learner of any other language) to manage is finding/paying* for the opportunity to talk to a native French speaker. The issue here isn't factual accuracy, but the ability to practice conversation (and hopefully have your grammar/syntax gently corrected.) I hate to say it, but AI is perfect for this task - who cares if the books it cites really exist or whether it thinks a battleship can sail across Nevada as long as the AI speaks correctly and uses a strong, useful vocabulary?

Also, academia?

https://acoup.blog/2023/02/17/collections-on-chatgpt/

TL:DR, he didn't like it!

* Not everyone has the money to pay a foreign language school.

As related to a discussion of AI, how have I spent the last two years never hearing about the Torment Nexus?

Sci-Fi Author: In my book I invented the Torment Nexus as a cautionary tale

Tech Company: At long last, we have created the Torment Nexus from classic sci-fi novel Don't Create The Torment Nexus

Beware the JBIG2 problem...

Troutwaxer @ 59: Actually, you've got a real point there. One of the hardest things for a non-French speaker (or the learner of any other language) to manage is finding/paying for the opportunity to talk to a native French speaker. The issue here isn't factual accuracy, but the ability to practice conversation (and hopefully have your grammar/syntax gently corrected.)*

I gave that a go just now on ChatGPT's trial:

Critique my grammer: "I were going downtown yesterday, when I saw a friend in the subway."

There are a few grammatical errors in the sentence you provided. Here's a corrected version:

"I was going downtown yesterday when I saw a friend in the subway."

The corrections made are:

"were" should be changed to "was" to agree with the singular subject "I."

The comma after "yesterday" is unnecessary and can be removed.

The word "in" should be added after "saw" to clarify that the friend was seen in the subway.

With these changes, the sentence is grammatically correct.

The first two points are correct. The last is ... not even wrong.

Maybe not so good for a language learner, but probably fixable. (Maybe it tries to follow a rule of three?)

This is stuff Facebook's AI research orgs have actually been working on. https://ai.facebook.com/blog/ai-powered-audio-compression-technique/ is the audio version.

They call the video version Codec Avatars https://research.facebook.com/publications/pixel-codec-avatars/ The original video source appears to have bitrotted, but googling find a fair number of folks who have duplicated all or part of it in their own videos.

The video version is targeted at taking data from a bunch of inward pointing cameras on a "Meta Quest" headset and turning it into a facial rendering, since the headset gets in the way of just pointing a camera at your face. Some interesting basic-ish research coming out of Zuckerberg's obsession with the Metaverse.

"Um - "The link on the orange site" - uh?"

A site which names itself "Hacker News" but confusingly uses the domain name ycombinator.com to make it difficult to search for, and has an orange strip across the top of the page. It seems that a lot of Charlie's articles get linked to from there.

Its discussion on this article has quite a few people saying they find this thing useful in their jobs. Which are... programming. To which I say... aaaarrggh, and "Oh for Cthulhu's sake", and other things of a like kind.

I've played with ChatGPT a little. On one hand, as far as factual content is concerned, it makes egregious errors (when I asked it to explain a line from one of Joni Mitchell's songs, it gave a lucid explanation of it as a line from AN ENTIRELY DIFFERENT SONG); it seems like simulation of a college freshman trying to bluff their way through an exam when they haven't done the reading. On the other, its style is no better than an unsubtle pastiche. I don't see it as being of any real use in generating content.

As a tabletop game master, I've played with a lot of systems for generating things through random rolls: interstellar sectors, solar systems, alien lifeforms, intelligent races, superheroes, dungeons . . . I've never seen one that produces usable output. At best you get things that have to be massively curated. It really seems like less work to rely on intelligent design in the first place. And ChatGPT seems mostly like a device for generating things through random rolls and table lookups, on a gargantuan scale, like the proverbial billion monkeys.

I hadn't thought about the economic implications, but your advice seems sound and I'm glad to have it.

That being said, I'd be extremely surprised if there wasn't already some form of steganographic info embedded into the various generators simply to identify their own generated content compared with an opposition system.

No need to be surprised, it already has been discussed. I feel as stupid as an AI - don’t remember where I saw it :-).

The technique was to divide the vocabulary (words and parts of words, about 50,000 items) randomly into a red list and green list. Then when picking the next word alternatively choose the highest probability word from one list, then the other list. If the generated text has alternated red and green words, then it’s very likely from that AI.

For extra subtlety and better word choices, allow words from the other list occasionally. You can still say the text is from that watermarked AI with ridiculously high probability.

Decent piece on chatGPT from Ted Chiang.

The point about the comma is debatable, too; it can be removed, but that doesn't mean it has to be, and whether or not it should be is undecidable without more context (which of course in this instance does not exist).

I'm dubious in any case about trusting anything trained by scarfing the internet on grammar and punctuation, given the amount of gruesome grammar and worse punctuation that is to be found even on supposedly "respectable" sites. It's not just binary "correctness", it's the ability of subtle changes in usage within the "correctness" envelope to alter tone and meaning, sometimes quite considerably. A lot of real people are not literate enough to be sensitive to this, and write things that don't make sense as a result (for numerous examples, see the current state of en.wikipedia articles about the Western Front in WW1, which have recently? been expanded by someone who knows a lot of military detail but doesn't have a clue about where commas ought to be).

I reckon the principal advantage in talking to a real person is regarding pronunciation, which should be obvious enough to need no expansion; and secondarily, real-world use of the language. The prim correctness and literalness of "schoolboy French" may be adequate to suppress criticism of the schools for "teaching kids wrong", but it's not how real French people speak or write, and they very likely aren't going to pronounce it the same as the teacher either.

So this article showed up. Its thesis is that the Ukraine War is pushing adoption of more and more independent AI weapons, one that can act without having a human in the decision loop. It's from The Conversation, so take that into account.

https://theconversation.com/war-in-ukraine-accelerates-global-drive-toward-killer-robots-198725

My thought was a scenario: some unnervingly Skynet-like system kills a bunch of people it wasn't exactly supposed to kill. When queried about why it made that decision, its software directs it to engage ChatGPT with some set of predetermined prompts.

Unfortunately, I can see militarized bullshit generation as a niche, but profitable, market for these kinds of systems.

Speaking of iterative bullshit generation, inquiring minds want to know:

--How are chatbots at producing VC pitches?

--Are chatbots really good enough to help someone get an MBA by writing their essays for them? ( https://knowledge.wharton.upenn.edu/podcast/wharton-business-daily-podcast/chatgpt-passed-an-mba-exam-whats-next/ )

--Has someone checked to see if speeches by MAGAt political figures, including Trump, pass or fail the test to see if they're AI generated? What about other politicians? The point here is to check if human-generated iterative bullshit is different than that created by AI).

--If politicians use chatbots to produce scripts for speeches, how good is AI at speechifying? I can easily see lower-level candidates using AI to write speeches for them.

--As a follow-on, are chatbots better at emitting left-wing speeches, right-wing speeches, or are they equally good at bullshitting across the political spectrum?

Help on any of these?

Language isn't sufficient for intelligence, and it's not required for intelligence, though it's required for shared intelligence.

OTOH, Microsoft has announced (how seriously?) that it's working on connecting ChatGPT with a robot body. I think that has the true potential for intelligence. I feel that they should solve the "asserting fictions as truth" problem first, but I'm not certain. Perhaps "living in the world" is necessary to be able to distinguish between them.

That said, WRT the basic point, AI stocks are probably overhyped right now. The fact that I'm convinced some real change is happening doesn't deny that claim at all. It's quite possible that none of the companies that we've heard about will be the real winner here, but another change is in process. It's not clear what the "lead product" will be. "Script writer" is a rather niche occupation. Maybe it will be a market researcher. A "Tell me what you want and I'll recommend a product" application. Consider (going back to an earlier discussion) "I'm 6 feet tall, and I've got knee problems. My wife an I want airplane tickets from Silver City, Nevada to Athens, Greece. What airline tickets should I buy?" or "I want a blue suede couch to match my blue suede shoes. What are my choices?". This won't work if it's going to fantasize the answers, but it something a ChatGPT style application might do quite well at.

It's my guess that prior to ChatGPT there wasn't any real motivation to have conclusions gradient with respect to fact for a generated speech engine. I know I've seen statements that "The reason ChatGPT doesn't reliable give true statements is that that isn't what it was trained to do." That's supposed to have come from somebody knowledgeable, but I don't know the validity of that claim.

Clearly if you're going to be training your AI with data from the internet, requiring that the data is true as well as requiring that the data is intelligible is adding a high bar. So I tend to believe that the above quite is correct, even if it's been paraphrased a bit (possibly by me).

That's a mix of very valid and somewhat questionable points. It's certainly true that much thought is non-verbal. You don't need to go to athletes to show that, just try to describe in detail how you get out of a chair.

OTOH, the claim that human thought isn't different runs afoul of the FOXP2 gene. People have a different form of a really highly conserved gene that is active in the brain, and when it gets mutated, the ability to use grammar is pretty much killed. There are also a number of other genes affecting how the brain works that differ between humans and other apes. I have a strong suspicion that these affect things that I call (without evidence) "stack depth". I.e. the number of past events that one can track at the same time, like tracking what a pronoun refers to over a couple of paragraphs. (I think my label reflects an invalid model of how thought works, but it labels a real phenomenon.)

That said, (and IIUC) people with a mutated FOXP2 gene can still dress themselves, and safely cross a street at a traffic light. So it doesn't affect the entire process of thought. (I couldn't validate this with a quick search, but I think it's true. The search did show, however, that the mutated FOXP2 gene also made learning sign language impossible.)

Project Gutenberg would be a good source of public domain text for that approach. All the works there are public domain. But it's for story (mainly book) length fiction. The dialogs aren't proper for interactive speech.

Perhaps what they need to do it start a competitor to rededit with a license that puts everything posted in public domain. With a bit (not too much) of advertising it might even pay for itself. And you could have community moderators who would rate the posts into useful categories. Mix that in with the language learned from the public domain books and you might have something.

The problem is that that's what people do, too. There needs to be a "smallest degree of similarity that counts as a copyright violation", and once you have that, you'll find famous authors sued for violating it by accident. Currently that safe because there's no well defined boundary, and there's no automated "search for violations", but once those exist to deter auto-generated text, they'll also work on human generated text. (See "My Sweet Lord". But that's just a famous example.)

See Spider Robinson's "The Melancholy Elephants".

It's not just the VC's. I have a friend who is really convinced that an AI able to handle go has intelligence well above the average for humans.

That said, you CAN generalize from that to many similar tasks. But to be able to learn several different games without forgetting the rules of the earlier ones I found more impressive. (OTOH, I'm still estimating the Singularity at 2035. But I expect lots of shouting and flashy stuff along the way.)

OTOH, there are signs that I may have posted that data a bit too late, but I'm not convinced. I think there will be unexpected problems, and possibly even 2035 is too soon. But you can't really tell. We don't understand the problems, and it could be just one small subroutine away from something really impressive. That's not the way I'd bet, though.

72 Para 3 - You need an answer to a different and more detailed question, such as "I am 6 feet tall with a 34 inch inside leg. Which airlines fly from Nevada or California to Greece, and have a suitable seat pitch for me to get comfortable?"
Now your expert system can run a search and say something like "Business class seating on {list of flag carriers} operates on your routings and matches or exceeds your required seat pitch based on your height and inside leg. No Veal Crate class seating matches this requirement."

Language isn't sufficient for intelligence, and it's not required for intelligence, though it's required for shared intelligence.

Surely the basic requirement for shared intelligence is just observing the behavior of others and remembering it later.

“ No. What happened with "the cloud" is that it shifted CPU cycles from the capital expenditure side of the balance sheet to an operating cost.”

Sure that’s literally what I said

Renting is often (dare I say usually) a better deal then buying for an early stage startup doing ML based work, based on the typical ML usage patterns. The majority of the compute workload is generally training the model which usually only happens every now and then. It goes “small scale training on a single machine (Ml dev basically), ok we think we want to do a big training run, on it looks good let’s run inference “Once you have a model trained, running inference, is lightweight.

Buying gpu’s would be like buying a moving van because you are planning on moving . Sure the guy renting you the moving van is charging you a markup but that’s better then the moving van sitting in your driveway till the next time you move.

As a real world example, when we do a full training run in aws we typically run around 100 g5g.2xlarge, each with a 16gig gpu. We rent that cluster maybe one e ever 2-3 weeks for aboit 3-4 hours

There needs to be a "smallest degree of similarity that counts as a copyright violation", and once you have that, you'll find famous authors sued for violating it by accident.

Luckily the US Supreme Court recently ruled that there is no copyright on content generated by AI (that is, the current stuff called "AI" in public discourse). So you can't simply generate the set of all possible (or probable) Charlie Stross novels and then sue me for copyright violation/plagiarism.

FOX-P2 Zombies?

Yeah, now imagine you're running that cluster 24x365. Or worse, needing 1000x VMs, not 100x. Sooner or later it's going to be cheaper to buy your own warehouse.

-Are chatbots really good enough to help someone get an MBA by writing their essays for them?

Years ago I read a book by a chap who made a living writing essays for other people*. In one case he earned a Masters for someone by doing all their writing, including their thesis. The latter was written in one incredible cocaine-fueled allnight session. (OK, several days non-stop.) It was, in the writers words, bullshit, but sufficiently convincing bullshit to earn the customer his masters degree, which was all he cared about.

If the bar is "cocaine-fueled bullshit artist with no sleep" then the answer to your question is almost certainly "probably".

*His rationale was that while he was at university the only department that actually cared about students was parking enforcement. Books missing from library, relative died, professor not available, lost in classes of 500+… no response, but park for five minutes without the correct tag and by dog did you university insist on contacting you. So as the university clearly viewed his degree as a transactional affair, why shouldn't he view someone else earning a degree the same way?

If you find yourself running it 24/7 you need to buy your own hardware

However that is not a thing that is inevitable or even very likely to happen. Not more likely then you find yourself moving houses every week.

If you need 1000 machines you need to make a call to your Amazon account rep and let them know that is coming ahead of time.

Saw this tonight from Steve Jackson Games:

http://www.sjgames.com/ill/a/2023-02-26 Andrew Walters writes: "I am asking ChatGPT questions about Munchkin rules and cards. It is wrong 100% of the time, even with questions that are on the FAQ. ChatGPT says, explicitly, that the Munchkin FAQ was part of its training material.

"The responses were perfectly grammatically correct, detailed, and sounded very confident. And wrong every time. Well, 0/5. I have a transcript."

More in the complete post.

OTOH, the claim that human thought isn't different runs afoul of the FOXP2 gene.

The question is different in kind or degree. You're arguing kind, I'm arguing degree. FoxP2 genes show up in everything from fish on up, and disabling the gene always results in neural abnormalities in cases where the experiment was done. In some groups it's conserved, in others like bats, it's highly diversified and apparently related to echolocation ( https://en.wikipedia.org/wiki/FOXP2 ).

Regardless, other species have demonstrated understanding hundreds of words (e.g. border collies) and sometimes vocalizing them (Alex the parrot). That's why I'd argue human intelligence differs in degree, not kind, with other animals, and that intelligence isn't limited to manipulation of symbols in any organism.

Speaking as a professional copy editor, who mostly works on manuscripts whose authors have doctorates, I'd say two things about Pigeon's comment 69: On one hand, very few of those authors have mastered the subtleties of punctuation, or of syntactic issues such as agreement and antecedence. On the other, it's nearly always possible to figure out where the commas ought to go. This has inspired what's called "open punctuation," an approach that says that if you can tell a comma belong in a sentence, then you don't actually need to put the comma there . . .

now imagine you're running that cluster 24x365

For generic compute resource, sure. But lots of people actually want less than ~6 hours of downtime a year, including connectivity. So it's "24x365.24-ish compute + connectivity" and that's hard to get out of a warehouse full of computers.

If you want reliability it's common that you need good connectivity in three or more locations in different geographic areas. So the step from someone else's computer can easily be a significant international operation. When we made the leap from "a couple of servers on top of a filing cabinet in an office in Europe" to the cloud it made a whole lot of stuff easier. And meant that random internet outages stopped being a problem (... were less of a problem)

We did go through some learning experiences with what exactly "uptime" means (eg, ~60s of downtime while they moved a virtual machine to a new host doesn't count as downtime). OTOH same deal with our office, the ISP's "SIM backup" went down when their fibre did... ooops)

(also, Joey says "don't touch these little fine hairy spikes like I just did")

Not being a pedant I haven't done a deep dive on the subject but I understand the use of commas is one of those US/RoW things that divide the English-speaking nations. Strunk & White supposedly differs from the Oxford Comma school and so a lot of American writing is littered with commas that other English writing would not have.

https://lite.cnn.com/2023/02/26/us/nyc-airpod-max-theft-trnd

"Moped-riding thieves stealing AirPod Max headphones off people’s heads"

what surprises me -- the only aspect about this crime -- is how long it has taken criminals to adapt behavior to maximize their haul of high value items whilst minimizing their risk...

which is an extension of the opening premise of this thread... market-driven evolution by grifters seeking that sweetest of 'sweet spot' of maximize their haul whilst minimizing their risk

not enough to find something of value to steal, it is important to get away with it...

crypto-currency having just about reached peak grift and now either forced to go legit within explicit legislation to join mainstream capital markets (FX, commodities, bonds, etc) or otherwise driven into the shadows by regulators

Sometimes the VCs and vendors don't even need PR - they just publish a white paper or blog post that's somehow accepted as credible, then sit back and enjoy the floods of credulous cash and FOMO-fuelled followers. Then the press chimes in with 'new buzzword is the future'. The VCs are in and out before the rest of us get to wear the losses, and the wider damage.

Yes, that's why I think of cats as small, furry, feature reduced people.

It's just a new target for an old technique. Moped-riding handbag snatchers have been around for ages.

It looks to me like the cryptogrifters these days are into combining it with that ancient and traditional grift, the gambling industry. At least such is the impression I get from spurious results thrown up while trying to find the "orange site link" to this article and getting deep into the cack results through not realising that said site's domain name does not match its title.

You're welcome to think it's all a grift, but I asked it to write some very specific code, and .... well, you can read for yourself:

https://twitter.com/russnelson/status/1628960840069120004

TLDR: Summarizing, it felt like I was guiding a noob coder on how to convert a test from fudge to mock. It produced the same code that I wrote by hand (at the tweet ending in ....), so not bad. Of course, it's very simple code, a bare minimum of a test. Still, it was very entertaining.

For clarity, this means "Steve (US) Jackson", not "Steve (UK) Jackson".

Strunk and White are wrong about almost everything.

Given a choice between burning cycles on AI training vs. crypto, I'll take AI training every time. At least with AI, work is going to produce something with actual value, as compared to crypto cycles that have no intrinsic value whatsoever. (Hell, you can burn pengoes for warmth, at least.)

Every time I see "AI" I think of sloths. It's far more relevant, since at this point it's mainly a marketing term.

Unholyguy @ 16:

VC money is dumb and herd like and will certainly pile on the sexy trends.

As stupid as they are with it, I'm kinda wondering how THEY acquired it in the first place?

This is a very interesting discussion, but this comment section seems to be quite bearish on AI/ML. So I want to add my two cents:

• As a counterpoint to the position that AI is like crypto: John Carmack has recently switched to full time working on AI/ML and that alone makes this "less like crypto".
• As a language model, using ChatGPT to reproduce/recall factual information accurately, is a weird form of critiquing it. Things I have used ChatGPT for:
  • Create items and villains for a RPG, including stats, backstory, and motivations
  • Run a choose your own adventure-style story with ChatGPT as DM for my five-year-old (if anybody wants the "engineered prompts", tell me. They're fun)
  • What I also find very impressive is its ability to do that in German, French, and English (and probably other languages, those are just the ones that I speak). I also used it to translate, and then slightly edit, contractual language recently in English and German.

It definitely did a good job at the above. For the RPG stuff at least on the level of when yours truly was DMing in high school. So I can easily see a path of iterative improvement where something like ChatGPT could power a MMO and you can bet that it has uses in erotic entertainment. The way that I think about its literary prowess is exactly this: ChatGPT's quality is at the level of "ask a really confident high school student to write a fantasy novel or to try to bullshit their way through class". And for that, it works really well.

This is also very different from the cryptosphere where a clearly visible iterative path to profitable adoption is missing ;).

I finally figured out what this whole bespoke bullshit generation thing reminded me of: Douglas Adams's Dirk Gently's Holistic Detective Agency from 1987.

"Well, Gordon's great insight was to design a program which allowed you to specify in advance what decision you wished it to reach, and only then to give it all the facts. The program's task, which it was able to accomplish with consummate ease, was simply to construct a plausible series of logical-sounding steps to connect the premises with the conclusion."

Pigeon:

not the technique it was the target that had gone un-stolen for a rather stunningly long time... nobody grabs rings off of fingers but necklaces are always an obvious and fast target... likewise something that costs USD $400+ and requires merest tugging upon to steal...

my contribution this day to the thread was merely snarking upon how there's shifting in what snake oil is getting bottled as well a re-design of the packaging... but as always there's real attempts at 'next big thing' leading into 'newest big grift'... somehow all this AI crap will lead into porn (visual not textual)... still one of the few profitable products on the internet...

problem is how can I wrap a pitch in almost-plausible bullshit to draw in the VC money and distract the CPAs from digging too deep whilst wining 'n dining the MBAs...

...heh heh

excuse while I register "pornGPT" as a trademark prior to pitching my grift... and ditto the website...

I can see there's gonna' be a lot of new words & expressions to look up (new to me anyway). I like that. It's always fun to find out new things.

One of the tangents I discovered from one of the links I followed suggests the reason "modern" humans displaced the Neanderthals is the Neanderthals didn't keep dogs, and "modern" humans did.

Still leaves the question, "Who domesticated whom?"

In the latest episode of Last Week Tonight, John Oliver discusses AI issues.

"Moped-riding thieves stealing AirPod Max headphones off people’s heads"

Huh. I have a pair of Airpod Max and I wouldn't dream of wearing them out of the house -- they're very distinctive (and distinctively expensive). They're also absolutely not rainproof (and expensive)! And the noise cancellation is sufficiently good that I'd be afraid of walking under a fire engine or ambulance despite lights and sirens.

(I keep hoping Apple will release a second gen set of wireless cans that are cheaper, smaller, less obvious, and more moisture-proof -- like the Homepod Mini compared to the full fat Homepod -- but it's been a couple of years now and, welp, guess I'll have to stick to my in-ear phones for walking-around music. Which I need, because despite all the other good things about Edinburgh, the stone canyons make it very noisy indeed at street level.)

So what is AI then? It's obligate and incredibly sophisticated symbol manipulation, using a system that uses symbols to process other symbols. In other words, it looks a lot like the way a believer might conceive The Word making reality. However, that model might have fairly little to do with the kinds of information processing life's done for the past two billion years. We'll see how it does as it hits reality.

That happens to be incorrect. One of the discoveries/inventions that made things like Bert and all the GPTs possible is embeddings. Which are associating (projecting is often the term) to each input symbol a vector of real (well, float) values of fixed size (usually 300-500). The model then works in the space of numbers on those values and do unspeakable things to them. Generating words at the end is usually though a one-layer network that decides which is the most probable symbol given a vector.

Which means those systems very very much do not manipulate symbols, but numbers. Which is also what makes GPUs relevant. And also that is one of the reasons it is so hard to even start to understand how they actually work once trained, the numbers are very, very opaque. But it "feels" closer to intuitive understanding/reactions than symbolic reasoning to put that back in anthropomorphic terms.

"*His rationale was that while he was at university the only department that actually cared about students was parking enforcement."

You made my day here. Thanks for the laugh.

Compare and contrast:

ChatGPT, which cost millions of dollars and thousands of man-years to develop (including all the research that led to it), is outputting boring but plausible text. Everybody is losing their shit.

Dwarf Fortress, which is made by a programmer and his brother over the course of ~20 years, is generating amazing narratives. Everybody is praising it to the hilt.

My bet is on Dwarf Fortress going full Skynet while humanity is distracted with neural networks.

For added shits and giggles, Ace (the imprint of Penguin, pre Random Penguin merger chaos) used to have copy edit guidelines that specified using The Chicago Manual of Style as a guide to grammar/construction/punctuation etc. in fiction.

Yeah, right.

This led to me writing "stet" in the margins a lot (back then it was "check your copy edits on paper") and fuming. I will note that my current editors at Tor agree with me about the wisdom (or lack thereof) of prescribing a guide to academic style for editing fiction ...

I can see James Joyce submitting 'Finnegan's Wake' and then some editor returning the manuscript, commented by The Chicago Manual of Style.

"While I agree with you on the hype train, I think there is also something swelling underneath."

Yes and no.

As Bret Devereaux said in his blog ( https://acoup.blog/2023/02/17/collections-on-chatgpt/ ), we (humans) are very very good at Anthropomorphism. Think for instance that there is a human statue of Justice (a very abstract concept) above most XIXth century tribunals or that we often attribute human motives to our cats and dogs.

Seeing a model that spew syntactically and grammatically mostly correct English, we immediately attribute to it a lot more understanding than it has really.

The progress is real, and it may be useful to put flavor around a more sophisticated system, or to help people who have trouble expressing themselves re-frame their words (providing they are experts on the subject and can filter out the nonsense). But right know the system understands absolutely nothing, and brings no understanding to its users. A similar situation is the recent win by an amateur human player against a Go playing AI. To win, he used patently absurd tactics that would not have fooled a human player, but were not in the training set. The AI (strong enough to beat the best professional players) has no real understanding of the goals and is thus very vulnerable to adversarial attacks.

What it does right now, in my opinion, is massive over-fitting (in the physicist sense) on a large dataset. It brings no understanding to the table. It can be good enough to pilot some industrial systems within relatively narrow parameters, but will spew completely absurd results when confronted to the unexpected.

Also I think there might even be some benefit for near-future sf from the fancy Markov Chain generators: the plausible-sounding complete bullshit generator can be a plot-element, useful for example highlighting the problems inherent in trusting them in any kind of real world scenario (excepting ones which rely on plausible-sounding bullshit anyway).

One of the tangents I discovered from one of the links I followed suggests the reason "modern" humans displaced the Neanderthals is the Neanderthals didn't keep dogs, and "modern" humans did. Still leaves the question, "Who domesticated whom?"

ISTR Alice Roberts talking about this question. Her book, Tamed: Ten Species that Changed our World, covers humans taming dogs and other species.

I've also heard her claim that cats domesticated themselves. ;-)

I immediately thought, "Like Power Point!" I may be too cynical for this poor, innocent world.

Martin Rogers
Cats: ... "Free FOOD!" - if we make nice with the Humans & purr a lot - AND we get tummy-tickles!

Every time I see "AI" I think of sloths.

I first encountered the acronym while working for the Department of Agriculture, back when personal computers were incredibly uncommon and modems were acoustically coupled.

So yeah, my first thought when I see a headline with "AI" is "what a load of bull****" (where the asterisks are not standing in for feces).

That happens to be incorrect.

We actually agree, we're just using terms differently. To me, symbols include numbers and words coded as numbers.

Thinking without words is basically doing everything by feel.

“As stupid as they are with it, I'm kinda wondering how THEY acquired it in the first place?”

It’s an interesting question. I think the answer is two fold 1: the stock market is also a stupid herd beast so as long as they can influence thar beast they can pull off a good ipo, they make money and get investors 2: overall the tech sector has been incredibly successful at producing value over the last thirty odd years. A rising tide floats all boats.

ISTR Alice Roberts talking about this question. Her book, Tamed: Ten Species that Changed our World, covers humans taming dogs and other species.

Domestication is our special word for mutualistic coevolution, when it occurs with humans. Theorists about it tend to fall into the pit of humans being different and special, and domestication is done by human will. If you get rid of the "humans being special" and "will matters" ideas, then we're again different from other species in degree, not kind. Oaks and pines mutualistically coevolve with dozens to hundreds (thousands?) of fungi, for example, and depend on them for everything from pathogen protection to mineral nutrition. Argentine ants routinely "domesticate" local aphid and sapsucking bug species. This makes them a real freaking nuisance in the garden, where they protect bugs that infest plants and transmit plant diseases in return for the surplus sugar the bugs poop out.

The thing is that species also cooperate and coevolve without becoming domesticated. For example, my mom's fed her local ravens for about 25 years now, and they warn her when things concern them. Some of them know me too, and even though I live over 100 miles away, I recently had a raven sit on a phone pole over my head and give me the contact call my mom developed (I responded, of course. My wife was bemused). Those ravens are not domesticated, but we're in a cooperative relationship. The old myths about how, in ye olden tymes, people and animals talked probably refers to relationships like these.

Neanderthals probably didn't domesticate dogs, but it's quite likely that if they stayed in an area for awhile, they knew who the individual local wolves were, and worked out some form of coexistence with them. This sounds all mystical until you think about it and realize that (to use modern phrasing) predators tend to not want to get into unnecessary squabbles, and big predators occur in low enough numbers that relating to individuals is entirely possible. It's actually not that hard to do it now, if you pay attention, as my mom's done with the local ravens.

Yeah, maybe Alice Roberts is right. ;)

I also recall Alice Roberts talking about coevolution between hominin species. The BBC documentary where she showed the genelines intersecting sometimes gets repeated. TL;DW Our ancestors and their 'cousins' were really friendly, and we have evidence for this in our DNA.

"Thinking without words is basically doing everything by feel."

1) Er - do what?

2) Often when you are forced to do something by feel that you would normally do by sight, it's not the doing it by feel per se that makes it difficult, it's the (often unrecognised) need to stop trying to do it by sight.

3) Lots of things are mostly done by feel anyway. I expect you can tie your shoelaces perfectly well in pitch darkness, if you've ever tried it, but if you've had your hands in a bowl of ice for ten minutes beforehand then even full noon sunlight doesn't stop it being difficult.

"relating to individuals is entirely possible. It's actually not that hard to do it now, if you pay attention"

Seems to me that the biggest difficulty to overcome is figuring out how to tell the individuals apart, getting past "they all look the same". They don't, but the points which distinguish one raven from another have so little in common with the points that distinguish one human from another that it's too easy just to assume there aren't any and not bother trying to look for them. Once you do cotton on to what to look for the rest just follows.

New! { old, re-found } Pterry stories! ...

In other news, it appears that Sunak has outmanoeuvred the total-ultra Brexshiteers ... the actual "dead-give-away" came at the end of MS von der Leyen's speech: "This Windsor Framework is good news for scientists and researchers in the UK and EU," - She was, of course, referring to the Horizon Europe programme.
For that, the various headbangers can go & stuff themselves.

H
"Hugin & Munin" ?

Martin Rodgers
The term for that is Rishathra

Pigeon
The late Sir Peter Scott & his daughter realised that the bill-patterns on Bewick's Swans were unique & could be used to identify individual birds, & Polly Pullar in Arndamurchan has similarly realised that the "shirt-front" patterns on Pine Martens are also unique - I've wondered about the same type of patterning on Common Frogs, as well .....

Bugger, done it again: Ardnamurchan

Thank you, like blarney engines a lot, and intend to use it.

As I have been saying for many, many years - and I will happily add VC to it, to paraphrase the Batman, stock traders (and VC) are a superstitious and cowardly lot."

I would not say it involves symbol manipulation, and suggest that's an artifact of symbol manipulators.

We know, and have observed, more than one chimp, in the wild, carrying a stick of the right shape and size, klick to get honey out of a hive, and in checking just now, https://link.springer.com/article/10.1007/BF02557584 immature gorillas using sticks and rocks.

What this tells me is that they, like humans, see something, and can ->generalize<-, that A can be used for obtaining B. The generalization of one object to another is what's significant.

Humans, and some other critters, are also time-binders: we teach kids how do use the generalized tools.

Thank you, but that phrase, "artificial stupid" is copyright KD and mark roth-whitworth, I think around '95 (before Flynn). She used it in a story in a fiction apa, so it was, effectively, published.

And the artificial stupids are named George, by default, because the fictional company that started creating them used the slogan "Let George do it".

I think if I had the enthusiasm, this would be the time for me to start that company in the real world. First example usage: George, filter my email. George: right, I know this is spam (you're not interested in twentysomething girls from Moscow who want to meet you), that's not, it's from Worldcon, and, um, hey boss, I don't know what to do with this, what do you want I should do with it, and should I do that in the future as well?"

In other words, it does NOT try to be "helpful" and guess what you want. If it doesn't know, it throws it to you.

Actually, I could get a perl program as a start: back in '93/'94, when I used to hang out on usenet on alt.pagan, we finally gave in, after being attacked by a bunch who went around usenet trying to destroy newsgroups, and created a big eight newsgroup, soc.religion.paganism. After months of screaming by some people as we were preparing the proposal, one of our people modified a program from, I think it was talk.russian, and we adopted his mod: if it was a regular poster, it was approved, a new poster, if it didn't see anythign objectionale (like YOU'LL BRUN IN HELL!!!), it approved it... and if it wasn't sure, it would randomly choose one of our four modkin (human moderators) to toss it to, and let them decide.

And that's how an AS should work.

Charlie's dead right here. Renting time, if it's not for a trial, is a waste of money, just like outsourcing to a contracting company... for a job that is ongoing, rather than short term.

I'll note that when my son was in distance learning to finish high school (let's not get into a racist school system in Chicago), for language, he asked to do something that was not overly expensive - and went to L'Alliance Francais for three months - immersive French.

And then there's other, older uses. Sam Johnson's dictionary of English - the first, in the 1750s, had commas used as pauses, as though speaking aloud.

Grump. My novel that was published would have been, as I noted in the acknowledgements, thousands of characters longer, had not a Certain Beta Reader not jumped all over me when I wanted to do that.

Don't be silly - they have the Big Picture in mind, and hired us to deal with the fiddly details.

You should warn people before you say things like that. It's just lucky that I'd finished my tea before I read it....

No, cats domesticated us.

Come on, who pays for the cat food?

Our ancestors and their 'cousins' were really friendly

Steady, there! I think the word you're looking for is probably not friendly but rapey (and there were survivors who carried to term and didn't abandon their weird-looking babies).

SMBC recently had a comic about countering toxic conspiracy theories by flooding the zone with plausible and utterly silly theories of equal value.

Someone posts a rant about Jewish bankers secretly coordinating daycare services for nefarious purposes, and they are immediately drowned out by a flood of Chatgpt generated plausible conspiracies that are equally stupid but also funny.

Result from a request to 'Write 12 20 word silly conspiracy theories'.

"Aliens are hiding among us as houseplants, waiting to take over the world.

The government is hiding a cure for the common cold to keep the tissue industry afloat.

Bigfoot is actually a super intelligent being and is hiding from us to avoid paparazzi.

The moon landing was faked because the government didn't want us to see the moon's secret alien base.

The Bermuda Triangle is actually a portal to another dimension, controlled by the Illuminati.

The earth is hollow and home to an advanced civilization that doesn't want to be disturbed.

The sun is actually a giant disco ball, but the government is hiding it to prevent widespread dancing.

The Loch Ness Monster is actually a government experiment gone wrong.

The earth is a giant snow globe, and earthquakes are caused by someone shaking it too hard.

The government is hiding a secret society of shape-shifting lizard people who control the world.

The pyramids were built by aliens to communicate with other civilizations through wifi signals.

The moon is slowly drifting away from the earth because it's tired of all the drama."

There may be something to the last one.

I felt obliged to make it a bit sillier, so here are 5 20 word silly conspiracy theories about breakfast cereal.

"Cereal is made with a special ingredient that makes us forget our dreams so we can't uncover their secrets."

"The shapes of the cereal are actually a code that reveals hidden messages from ancient civilizations."

"The marshmallows in Lucky Charms are actually tiny alien probes sent to monitor us."

"Breakfast cereal is actually a government mind-control device disguised as a tasty treat."

"The milk in our cereal is actually a secret formula designed to enhance our psychic abilities."

I seem to recall a plotline in Stephenson's 'FALL' that involves countering a gamergate-style hate driven conspiracy attack against a character by flooding the internet with gigatons of ever more stupid sounding nonsense about that character.

Steady, there! I think the word you're looking for is probably not friendly but rapey (and there were survivors who carried to term and didn't abandon their weird-looking babies).

Yah know, we talked about hominid interbreeding, from which we are ALL descended."

It really is toxic to assume that people who don't look like each other can only rape each other. I'm not sure why this is your default opinion, but speaking for all your fans who are in loving, long-term, interracial marriages, please drop it again.

Permanently.

It's just as ugly and unwarranted as anti-semitism.

*It's not just Africans, Neanderthals, and Denisovans. Our DNA also includes multiple "ghost lineages" from within Africa. Indonesians and Melanesians may also have DNA from late-surviving Homo erectus.

I have to agree. Sorry, forget the current name of "Esquimoes", but they had an active cultural thing of exobreeding, not just sharing food with a wandering Fuller Brush salesman, but also the woman as a sexual partner.

"then the battleship Iowa crossed Nevada."<\i>

This is entirely possible, if discussing naval maneuver, that the Iowa may have crossed the Nevada. They both served in the Pacific in WWII.

This is one of the reasons that natural language is tough.

What I'm getting at is (a) hominins presumably lived in tribes/troupes (like other primates), and (b) probably waged wars (again, like other primates -- eg. chimps). The question is how much peaceful inter-group exchange there was, as well as the hostile stuff. (IIRC fossil evidence suggests that interpersonal violence was a major cause of mortality among early humans -- indeed, it's only in the past few centuries that it's become less common among us).

Nojay: I've worked for a variety of US publishers for years, and every one of them has used the Oxford or serial comma. I do have Strunk and White on my shelves, but I don't consult them professionally, and taking a glance at them now, I'm not seeing where they offer a rule for using or avoiding the serial comma.

Personally, not that my personal prefereces enter into it when I'm copy editing, I prefer the serial comma, because it often avoids ambiguity.

Well... I should mention that my late ex told me how, when she was in the Navy as a cadet, the Captain congratulated her on navigating their (WWII? era) aircraft carrier to the center of downtown Norfolk, VA.... (No, the ship didn't follow her directions.)

What I'm getting at is (a) hominins presumably lived in tribes/troupes (like other primates), and (b) probably waged wars (again, like other primates -- eg. chimps).

What you're doing is propagating toxic and obsolete.

Humans are equally related to bonobos, which have never been observed killing each other and rarely fight in the wild, and chimps, which are considerably more violent. We share unique stretches of DNA with BOTH OF THESE SPECIES. We are equally similar to both bonobos and chimps.

The toxic bullshit you're propagating is Raymond Dart's repeatedly disproven "killer ape" hypothesis, which was based on crap evidence and came about before anyone realized bonobos were a separate species and not chimps. Our common ancestors with both Pan species probably looked more like bonobos anyway. We can as easily be described as unusually heterosexual bonobos as we can killer chimps tamed by civilization.

Go read Frans De Waal's Different: Gender through the eyes of a primatologist* if you want to get up to speed. It will also help you argue for equal rights for people who aren't strictly cisgender and heteronormative. Yes, the issues are linked, and you're repeating the idea that sex with someone who's "not the type approved by the group" can only be violent and non-consensual.

Until you read de Waal's work, realize that the science hasn't supported your position for decades, but bigots love the kind of apologia for violence that you're promulgating here.

*https://www.amazon.com/Different-Gender-Through-Eyes-Primatologist-ebook/dp/B098TZB3R4/

We actually agree, we're just using terms differently. To me, symbols include numbers and words coded as numbers.

Thinking without words is basically doing everything by feel.

I'd argue the LLMs indeed do everything by feel, fsvo "feel". The numbers stop coding words after a very small number of layers, to become who knows what else.

Yes, I should've used quotes around the 'friendly' - but I was remembering a comment by a geneticist. (A rather drunk geneticist, if I recall correctly.) So perhaps there was some nuance missing.

How much physical evidence do we have for/against violent relations amoung our ancestors? Quite a lot, it seems.

Violence was widespread in early farming society, says new study https://phys.org/news/2023-01-violence-widespread-early-farming-society.html

"The study raises the question to why violence seems to have been so prevalent during this period. The most plausible explanation may be that the economic base of society had changed. With farming came inequality and those who fared less successfully appear at times to have engaged in raiding and collective violence as an alternative strategy for success, with the results now increasingly being recognized archaeologically," says Dr. Martin Smith.

Ok, how far back does this go? More than 12K years? Sure, there are weapons older than that, but they could've been used for hunting food. So how violent were our ancestors before farming?

Some of us are politely asking for more research. Fortunately, we're getting it.

Thank-you for mentioning Frans De Waal's work. I've heard him speak about this, and he has many interesting things to say on this subject.

Frans de Waal on chimpanzees (2016) https://www.bbc.co.uk/programmes/b07wt6bj

Frans de Waal on Culture and Gender in Primates (2018) https://podcasts.apple.com/ca/podcast/frans-de-waal-on-culture-and-gender-in-primates/id1406534739?i=1000558629877

Learning from apes, fish and wasps (2022) https://www.bbc.co.uk/programmes/m0017k6s

That last show was hosted by Adam Rutherford. He was the drunk geneticist I meantioned earlier (but that was on another podcast).

"Humans are equally related to bonobos, which have never been observed killing each other and rarely fight in the wild, and chimps, which are considerably more violent. We share unique stretches of DNA with BOTH OF THESE SPECIES. We are equally similar to both bonobos and chimps."

While I do not want to diss our bonobo brethren and sisters, nor F. DeWaal, it is just that some of what we know about early Bonobo behavior is more from observations (of then mostly juvenile bonobos) in Zoos and less from in the wild and these early observations still direct our focus. Mind you, I am not saying that bonobos behave like chimps or let alone humans, just that extrapolation is problematic and that it is hard even for scientists to shake a first impression. (Before I get misunderstood, different primate species clearly have different social structures and behaviours that range from despotism to actual communism*, and humans are IMHO actually not all that bad, after all we manage to live together in wider aggregates of millions and mostly are decent with each other)

*) Being cheeky here, the only primate species that tried communism failed, but at least we tried.

I on the other hand find the "serial" comma somewhat of a distraction when reading text, like tripping over a mislaid flagstone on the pavement (ObUS: sidewalk). See also "er" instead or "re" at the end of words, the invisible letter "u" and other American peculiarities.

The argument that cats domesticated us, and the evidence is that we feed them, seems to have things backward. Providing food is a means of control. Consider an old-fashioned company store: Was it the coal miners who domesticated the mine owners?

My favorite story by James H. Schmitz, "End of the Line," envisions a far future where eating food that doesn't come from state-run tissue cultures is a certain death from catastrophic immune reactions. Schmitz is pretty explicit about the political implications.

As someone who had an actual AI/A-Life job back from the 90s I can tell you it's very easy to make AI do something interesting. It's far, far harder to make it do something useful.

VC pitches only need the 'interesting' bit though.

I remember how AI got a lot of VC in the late 90s. The expert system "revolution" that was going to make professionals unemployed as expert systems replaced doctors and lawyers, etc. So much hype!

One of the best classes I did at uni was a graduate seminar with Douglas Hofstadter called "AI, Hope and Hype" in 1999. We took AI projects and compared what was being written about them in popular science press and/or congressional records and/or VC investment papers, vs an analysis of what that "AI" actually did.

It was shocking. Really shocking. And it changed how I read anything about science in the press.

Typical was the Carnegie Mellon "self-driving" vehicle: it drove over 99% of the way across the continental US coast-to-coast back in 1995!
But when you dug into it, it did the bit of driving that was "staying in a lane on the interstate at constant speed (or following the car in front), driving only on days with good weather". Such lane-following is a very cool technology, and it is great that 25 years later it has become common: but it's a very, very long way from being 99% of what you need to drive a car everywhere.

Expert systems AI for medical diagnosis was more of the same - it could give a good answer to easy cases, but couldn't tell whether a given problem it was given was an easy case or a hard one - which effectively meant it needed so much expert supervision it was useless.

Etc, etc, etc.

But the hype! Oh, the hype!

Typical was the Carnegie Mellon "self-driving" vehicle: it drove over 99% of the way across the continental US coast-to-coast back in 1995!

Yeah. And back then it was a van with the back of computer racks in the back.

But they kept working on it. Which is why Uber and I think Google opened research centers there. And started poaching profs and grad students.

I lived in the area in the 80s. When it was very much in it's infancy. But they knew at some point things would get small enough and cheap enough to make it practical.

Martin Rodgers @ 114:

I've also heard her claim that cats domesticated themselves. ;-)

I'm not all that sure cats ARE domesticated.

Cricket Explained: (stolen from another forum)

You have two sides; one out in the field and one in. Each man that's in the side that's in goes out and, when he's out, he comes in and the next man goes out to be in until he's out. When they are all out, the side that's out comes in and the side that's been in goes out and tries to get those coming in, out.

Sometimes you get men still in and not out.

When a man goes out to go in, the men who are out try to get him out and, when he is out, he goes in and the next man in goes out and goes in. There are two men called umpires who stay out all the time and they decide when the men who are in are out.

When both sides have been in and all the men have got out and both sides have been out twice after all the men have been in, including those who are not out, that is the end of the game.

Much simpler than the American game of Baseball.

I'm not all that sure cats ARE domesticated.

Ever meet a bobcat or a wild cat? Our kitties are very much domesticated little buggers.

The thing to realize is that genetics play a huge role in what domestic means, and every species is different.

Apparently some spiders can figure out how to get from A to B in a three-dimensional maze/obstacle course, in some considerably more efficient manner than a simple exhaustive physical exploration/search of the possibility space.

If this is true, then since they're not doing it by straight physical exploration, they must necessarily be forming some kind of abstraction of the maze to evaluate mentally. A map, in the general sense. Is it symbolic? Who the fuck knows, but there's nothing that says it has to be, and the principle of parsimony suggests strongly that it very likely isn't.

I reckon, too, that abstraction is a necessary foundation for generalisation, indeed one that's difficult to separate out from it. This covers the apes with pokey-pokey things. But again, symbols are not required.

Genetically, wildcats and domestic cats are close enough to interbreed. It happens sometimes in Scotland, where there are some. Indeed the kittens do tend to be right little savage shits. But feral domestic cat kittens that don't grow up habituated to human contact don't come out much different, especially after a couple of generations. The genetics still seem to be set pretty firmly at "savage", and it's very much environment that adds the various degrees of "...but will get on with humans if necessary" as a qualification.

But feral domestic cat kittens that don't grow up habituated to human contact don't come out much different, especially after a couple of generations.

My understanding with both domestic cats and dogs is they need to be around people very early or they tend think us us big folks as the enemy. Dogs need to be around people by 10 to 12 weeks, preferably 6 to 8. Not sure of cats.

Re: 'The belief that some (nearly all) people have, that there is some "there" there, inside silicon bullshit (e.g. SFReader@34, sorry, not attacking you, that's just the way your comments came across; I see the same sentiments everywhere, even from people who really should know better), is down to humans' embeddedness in the world and existence as products of evolution. Silicon bullshitters don't know that there is a world. Humans have trouble imagining there not being a world, or a mind, or any physical existence at all.'

Sorry - I don't think that I understand what you're saying. Not being snarky - it's just that I'm not a techie/scientist and some of your phrasing suggests math-based similes that I'm likely unfamiliar with*. My rant was directed at the lack of critical assessment/response to the hyping of a 'magic black box' that can do any and everything.

With luck maybe I can find John Oliver's AI segment on YT. He's got a good research staff and usually is able to communicate concepts in an understandable and entertaining way.

*Yep - ending a sentence with a preposition.

CharlesH @74: 'It's certainly true that much thought is non-verbal'

Agree.

Also, problem-solving organisms incl. humans get/use data from a variety of inputs (senses and interaction modalities) and then through some neural-muscular mechanism weigh and synthesize these disparate inputs to arrive at an answer. My understanding is that AIs have only one modality - numeric shaped data inputs. My question then is: How do you test these systems against each other?

Scientists are still discovering new aspects of how the brain (senses, cognitive processes, emotions, etc.) and rest of the human physiology work together - the gut-brain axis is a recent example.

Uncle Stinky @97: 'Strunk & White'

At one multinational corp I worked at, one VP (parachuted in from another 'English-speaking' country) got so ticked off with the then-common report writing style that he distributed a copy of S&W to every single person in all the depts that reported to him. Weird but true - I still have my copy.

Mikko Parviainen (he/him) @113: '... fancy Markov Chain generators: the plausible-sounding complete bullshit generator.'

Yeah - one way to make sure that the 'AI author' never comes up with a new idea that can stand on its own. (No emergence.)

Greg Tingey @124:: 'pterry stories'

Thanks! - I've just put this on my calendar to check with my local bookstore.

"Consider an old-fashioned company store: Was it the coal miners who domesticated the mine owners?"

That did manage to happen sometimes. The record usually gets written as So-and-so the great kind philanthropist mine-owner/ironmaster/mill-owner/whatever, but often a valid alternative description would be to say that he figured out which side his bread was really buttered and that life was actually easier with a workforce that was not continually crapped on.

A few years ago, and it could quite well have been on here, someone posted a link to a page on a newspaper website which had a gallery of images produced by applying forensic reconstruction techniques to Neanderthal fossils, using modern libraries for clothing, hairstyles etc.

Not one of them would you pick out of any crowd in London or Birmingham or anywhere else with a decently diverse population. Not even a little bit. They all just looked like thoroughly ordinary people. Indeed one of them was the exact spit of a mate of mine.

Heteromeles @ 156:

I'm not all that sure cats ARE domesticated.

Ever meet a bobcat or a wild cat? Our kitties are very much domesticated little buggers.

The thing to realize is that genetics play a huge role in what domestic means, and every species is different.

Have you ever actually lived with a cat? Have you no appreciation of the absurd?

hyperbole [ hahy-pur-buh-lee ]
noun Rhetoric.
     1• obvious and intentional exaggeration.
     2• an extravagant statement or figure of speech not intended to be taken literally, as “to wait an eternity.”

See Also: Humor (or "Humour" if you are so inclined.

The killer ape theory is pretty batshit but the idea that pre Neolithic revolution humans didn’t kill eachother pretty regularly is pretty batshit too, IMO. And they probably killed Neanderthals too. There is even some fossil record there. Whether Homo Sapiens killing Neanderthals was a major factor in their extinction as opposed to other reasons, like out competing them is harder to nail down.

My personal guess is we killed them all, but that’s just a guess.

Yes, we have sites where bodies show people died violently.

The problem is, they're a tiny percentage of the number of people who lived. As Tyson Yunkaporta pointed out in Sand Talk, normal funeral rites for Australian Aborigines (he is one) involved exposing the body to the elements and leaving the vicinity, which is generally easier than digging a big hole. If that's the norm, the surviving skeletons that were anomalously buried may not provide an accurate representation of how people died.

Anyway, the original point was that animals coevolve with humans, which has nothing to do with humans fucking animals.

Then someone mistook coevolution for meaning that H. sap. interbreeding with Neanderthals, Denisovans, and others to produce modern H. sap. sap., which is not only not even wrong, it's a total non sequitur from animal domestication.

Then Charlie insisted that the only way Homo sap. sap.'s gene parents could have bred was by rape.

Now you're arguing that yes, humans are violent. Apparently this is to support Charlie's contention that modern humans are the product of xenophobic sexual assaults, which in turn somehow brings violent xenophobic sex into the question of how domestic dogs and cats evolved from their ancestors?

Is there a point at which people discussing this start passing as sane human beings? I don't think even ChatGPT could produce something this implausible.

So let's talk about hominin sex and evolution.

First off, humans separated from the chimp/bonobo line 6-8 million years ago, but there's good genetic evidence that members of both lineages occasionally interbred for the next million years or so, much as coyotes and wolves (and dogs) do.

Xenophilia is a fine tradition in hominins, according to our genes.

Per ( https://en.wikipedia.org/wiki/Interbreeding_between_archaic_and_modern_humans ), humans interbred with Neanderthals between 65,000 and 47,000 years ago, although the text points out that skeletons with hybrid features as young as 24,000 years are known, and ancient genomes also suggest interbreeding perhaps to 37,000 years ago. Overall, we're talking about sex happening, far more than once, over possibly 20,000 years or more. That looks a lot like coexistence, no?

The interesting bit is modern human DNA, which shows no trace of known Neanderthal Y chromosomes (father) or mitochondria (mother), and shows relatively little Neanderthal DNA ON X-chromosomes. So one reasonable hypothesis is that humans and Neanderthals peacefully interbred, but their children tended to be less fertile (for example, the no Y/no mitochondria would happen if only the daughters of Neanderthal fathers and sap.sap. mothers were fertile). This is not a problem in a species where children have to be raised by bands, because it meant more helpers per breeder. Given Ice Age conditions, they were probably near carrying capacity anyway.

Over time, through disasters like the Phegraean Field megavolcano ca. 39,280 BC in Italy, Neanderthal populations were decimated, the survivors hooked up with moderns repopulating the area from Africa, and the Neanderthals were genetically swamped. It's also possible that 20,000 or so years later, the Neolithic agricultural revolution further favored sap. sap. genes over those of Neanderthals, and here we are, with our small brains, big chins, and supremacist ideologies.

No violence is needed. Just chance and genomics. This is just as possible as moderns going on a, erm, 20,000 year-long (?!) genocidal rampage and ending up as Neanderthal hybrids anyway.

154 - In UK law cats are still "wild animals". :-|
158 - And that's why.

While I don't think rape is the default interaction for hominin relationships you could consider the perspective of the 50% of females. According to WHO 1 in 3 contemporary women have had unwanted sexual interactions, so it is highly likely that also occurred in the past with at least the same frequency. So some of the cross matings were probably non-consensual, regardless of the difference in species. Likely the same or more as within H. sapiens. It has been established that when the "r" word is not used, a substantial minority of males themselves admit to coercive interactions with females. I think Charlie has the right of it to some degree, although NONE of us has anything more than speculation. There is no way to gain evidence to back our hypotheses, we come at our conclusions based on our own perspectives.

"Domesticated" Cats
At the same time "snowshoe" Lynx adults or Lynx kittens can easily be persuaded that humans are good to be around.
There's a animal rescue-&-mini-zoo in Derbyshire & they had an orphaned Lynx tom kitten - photograph of this monster, as an adult, lying across someone's sofa, as relaxed as possible.
Then there's THIS - or - This one being combed - the rumbling purr!
I also remember the vixen we had on our allotment plots, who, not content with doggie-bikkies from my hand, tried to steal the entire bag out of my trouser pocket, having previously tested my boot for edibility. She was lovely.

John S
As in the ONE RUN victory of NZ over England, yeaterday, you mean?

cats don't love us just because we feed them...

haven't you fools realized they've decided that life is not complete without an array of variously size cardboard boxes...?

my cats are long dead but my favorite moment was bringing something home -- microwave oven, laptop, 8 speed liquefying blender, etc -- and my cats refusing to get out of the way of the unpacking process because they both wanted to lay claim to the same box... they'd argue about it then reluctantly take turns before abandoning interest after a week or so... which was when it was possible unfold it and toss it out...

bizarro thing they always regarded shoe boxes as best thing...ever...

what happened whenever I tried to vacuum the floor still triggers the giggling fits... they loved being vac'd by Mister Hose... kinda-sorta creepy watching 'em just lay there soooooo happy being vac'd

Indeed one of them was the exact spit of a mate of mine.

They were clearly twin brothers separated by time travel.

(Maybe?)

the only primate species that tried communism failed, but at least we tried.

Hard disagree, here: communism as a form of social organization is ubiquitous among humans! It's just that it operates at the family or kin group level, attempts to scale it up above roughly Dunbar's number of participants generally fail (I suspect due to the toxic interaction between the need for management to coordinate activities and the attractiveness of managerial roles to sociopathic personalities -- "managerial" as in "political" here).

Thanks, book added to my notification list for price drops. I'll get to it in due course ...

one VP (parachuted in from another 'English-speaking' country) got so ticked off with the then-common report writing style that he distributed a copy of S&W to every single person in all the depts that reported to him

Somewhere I've still got my BNR style guide for writing documentation.

The tech writer who distributed them explained that it was more important that documentation be consistent than that it followed any particular standard, and no she wasn't interested in what your grade five teacher taught you about sentences.

Given that we were engineers from a wide variety of school systems and countries, most with the typical 'I'm an engineer so I know better than you' attitude, in retrospect I'm amazed at how well the policy succeeded.

I'm sure it's just a coincidence that training neural networks and mining cryptocurrencies are both applications that benefit from very large arrays of GPUs. As in, hundreds of thousands to millions of GPUs soaking up entire nations' worth of electricity. (If I recall correctly, the latest ChatGPT model was trained on a supercomputing cluster that turns nearly $2M of electricity a year into waste heat: and it took a couple of months of 100% usage.) And of course, AMD, Nvidia, Intel, and the usual suspects have never imagined paying a PR firm to talk up markets for their latest products.

I think this is sort of like arguing that since unicycles and cars both use wheels, cars are just a passing fad that will soon become obsolete.

GPUs are a versatile tool - any computer algorithm that can be parallelized benefits from them, and a lot more algorithms can be parallelized than you think. It's not surprising that they find use in many computer science applications. Some of them will have staying power, and some of them won't, but I don't see any reason why the failure of crypto proves that AI will be a flop as well.

Well, aside from the general prediction that VC money is going to contract because interest rates are going up, but that's not a claim about AI, just a claim about the amount of money looking for speculative ventures.

Just for the record, I believe your thinking on this is very, very sane. My own guess is that a combination of things happened. There were probably one-night-stands, love affairs, life-long-relationships, orgies and also rapes, with the last being a small fraction of all coitus occurring between various human subspecies.

The idea that evolution is "red in tooth and claw" badly misunderstands the concept, as does the idea that two similar subspecies, one of whom will eventually die off, think of each other as "rivals" rather than "the tribe next door" or "those poor ugly folk in the next valley," "our shaggy and ugly allies" or "desirable mating partners" or "those strange folk we see less and less who are so good with herbs."

Is there proof that thought requires words?

As anecdata, I had a fever many years ago in which I stopped thinking in words for almost a day. I didn't stop thinking: I was just entirely relying on things like mental maps to cogitate (that's my most overdeveloped mental skill anyway, it translates easily to all sorts of graph data structures). In so far as one cogitates at all with a nasty fever, that is.

It was fairly relaxing, or it would have been if I didn't fairly soon pick up on the fact that I wasn't using words to think in at all any more, and then that I couldn't even when I tried and couldn't even really remember memories involving words -- I could remember book plots, but not book quotes etc. Once I realised that it was fairly terrifying (would they come back? am I wordless forever?). But it didn't stop me thinking, and as that chain of thought suggests it definitely didn't stop me reasoning either. Except of course it wasn't really like that because it was wordless and it's a bit hard to demonstrate that in a textual comment form :)

Symbolic manipulation: not the be-all and end-all after all.

Case in point: the time Google asked me to come give a keynote at one of their open days. 20 minute talk, this time next week, in ... Palo Alto? "Yes, we can pay for your flights. Sorry there's no honorarium or hotel."

I know Google are physical presence lunatics, but surely they of all people could have managed a talk over the internet?! Talk about self-defeating.

(But then I suppose they think everyone lives within a short distance of Silicon Valley. No, sorry, we don't.)

FOXP2 is also active in the endothelium, the achilles tendon, various muscles, the gastric mucosa, the developing lung, and sperm (but that last can be discounted: almost everything is active there, sperm are weird and crazy), as well as lots of other places. Whatever FOXP2 does, it's not a grammar gene and is likely to be only one of ten thousand or more genes involved in the development and operation of the process of thought. Heck, it's a transcription factor, they usually end up tangled up in almost everything so it's not too surprising they're tangled up in this.

However... the claim that thought isn't different does indeed run afoul of the fact that basically nothing else has recursive grammar allied to vocalized symbolic representations like we do (songbirds have recursive grammar, but not symbolic representation: nothing else comes close except perhaps whales). And obviously this is inherited, and obviously the ability to learn it is a genetically encoded trait (witness the fact that it is universal among neurologically normal humans). But that doesn't mean FOXP2 is a thing that has any more effect on the process of thought than any of the other 1/3rd of the entire coding genome that gets expressed in the brain.

It is interesting that small mutations in it have the ability to knock out grammar so tightly, but this is at most telling us something interesting about the lack of compensatory pathways in whatever grammar-related aspect of neural development the pathway it is part of involves. Not having very many compensatory pathways in a new trait like building a brain that can do grammar is not terribly surprising: nor is the fact that most of its other expression sites keep on ticking after it's mutated, because they are mostly ancient traits with lots of alternative ways to get a working organ system out.

Or Tristram Shandy :-)

"Thinking without words is basically doing everything by feel. "

That is complete bollocks - indeed, it's close to discriminatory. There are many people who think non-verbally (and that includes not thinking in terms of conventional mathematical formulae) - I used to do so almost entirely, and still do to some extent. With me, when I was younger, symbolic thinking came in third place to visuospatial (i.e. patterns) and conceptual thinking.

Fer chrissake, how to you think that radically new mathematics was invented? The existing terminology didn't even cover the concepts, so they CAN'T have used existing symbols. What was done by most such people was to think of the concepts first and create new terminology second. Yes, it makes it damn hard to communicate with you obligate verbal thinkers, but that does NOT mean our thinking is just 'done by feel'.

I'm confused. Why are you trying to start an argument with someone who completely agrees with what you just wrote?

Because you posted the opposite!

Because you posted the opposite!

You misread what I posted. Like you, I'm very much not obligately verbal.

I think any opinion here is assuming some particular social structure, and I'm not sure that's valid. E.g.to what extent has humanity dropped the genetic underpinnings of chimpanzee sexual customs? If people live in small family troops (i.e. all those living together are related) exogamy can only be engaged in with strangers...and it's dangerous to wander around alone, but exogamy is highly beneficial to survival.

I don't think ANY modern human is a good model, because all modern humans live in a curated world, where wild predators have learned over evolutionary time that humans are likely to be dangerous.

So the question becomes "How do you engage in exogamy?". The travelling salesman is one approach. Mate exchanges at occasional meetups can work, but somehow those meetups have to be arranged. (There's evidence for meetups going back to the old stone age, but that's not very far in evolutionary time. Still, but the time they left traces they were pretty well established...which isn't surprising.)

The evidence for "military conflicts" between early societies is... well, difficult to interpret. There was lots of it, but it's not at all clear that is was "us vs. those guys that look different", it's quite as likely to have been arguments with the neighbors over rights to a hunting ground. There are definite signs that Cro Magnon and Neanderthal often occupied the same caves, but at this distance we can't really say they did so at the same time. Or that they didn't.

So. Rape as an explanation isn't impossible. It seems (to me) a bit unlikely, at least an anything very common. (I.e., it probably happened occasionally.) But lots of tribes throughout human history have managed exogamy via "bride stealing" (see the origin of the word "rape"), and the women were often active co-conspirators.

All that said, we don't really know what the social customs of the early humanish folk were. E.g. "Why did the carefully place a bunch of bear heads in a cave all facing North?" (I think it was North. A quick search didn't find a link.) So to assume that we really understand the customs that didn't leave a clear mark is ... not well founded.

FWIW, we didn't fail at communism, only at Communism. They're very different things. With communism we discovered that it doesn't scale very well, and the larger the group the more charismatic your leader needs to be. It works well at the "really small village" scale, if there isn't much external contact. About 5 families (perhaps more, but I don't know) can often manage it in the correct circumstances without a charismatic leader. And in the circumstances where it works, it improves life for everyone. In early US culture it was called "being neighborly".

I think you misinterpret what Schmitz was saying. IIUC he was not saying that being sensitive to impurities in food either was caused by or even enabled strong central government. He was implying that the ability to eat nearly anything was nearly a requirement for colonizing planets with their own life forms. This was mixed up with a very confused idea of how genetics works.

I really liked that story, but don't overread the background pieces.

If people are going to build it or design it, then symbols are required. Symbols are how we share info. They aren't all verbal (and I don't mean sonic) though. But the verbal ones can be transmitted from individual to individual without horrible amounts of lost information. (Compare digital duplication of video tapes vs. analog duplication)

So then the problem is mapping the symbols onto the implementation. (Well, actually it's always into the implementation...but that's picky.)

OTOH, evolutionary design doesn't require symbols. It can depend on features that the symbols being used don't include. There was an FPLA that was designed via evolutionary design (I forget what it did) that required a circuit that didn't have any symbolized connection to any other part of the design, but was required for it to operate. Turned out that there was an unmodeled capacitive linkage. So an expanded way of symbolizing things could handle it.

It's my guess that there will always be a possible symbolization, but that it will sometimes not be obvious how one needs to expand/adapt one's current symbolization to handle the "currently unmodeled" result.

Re: My understanding is that AIs have only one modality - numeric shaped data inputs.

That's both true and as accurate as saying that humans only have one modality -- chemical interaction.

The two statements are exactly as meaningful. And as true (or as false, if you prefer). And pretty homologous in their relationship, though for people the chemical interactions shape the structure in a way that the numbers don't for the AI. And if you want analog style noise/losses in your system, just use single precision floating point. Or half precision (fp16) if your system will support it.

The question is what you do with the info once you've got it. AIs are using a very simplified model compared to what humans (or chickens) use. We don't know what all the missing pieces are, or how important each one is. E.g. part of our neural system is used to handle blood pressure. That's probably not important to an AI, but evolution is full of very indirect interactions, so we can't REALLY be sure.

“The problem is, they're a tiny percentage of the number of people who lived.”

I really don’t think there is any plausible way to support that statement but I am very curious to see if you have an argument there.

It’s certainly not true of humans post Neolithic revolution where most estimates I’ve seen put the self inflicted death toll vey high, percentage wise

I think drawing a very firm line between “peaceful coexistence” and “20,000 year long rampages” is making the issue very black and white when it reality I am sure there was a good bit of both during the overlap time period.

But what we can say for certain is a species that had been around for roughly half a million years was extinct at the end of a relatively short overlap period. I find it pretty far fetched to believe that happened without significant violence. Especially given all the other branches of our close relations followed the same pattern. As did many megafauna. “Homo Sapiens moves into your territory, extinction follows” is a clear pattern for species with certain characteristics

Well, consider: we hung out with dogs for what, 20k? 30k? 50k years, and what we did was chase game, licked or scratched our private parts, and told stories. Then cats domesticated us 17k? 12k? years ago, and the next thing you know, we've got agriculture and civilization. See?

I'm willing to hear you admit that cats made agriculture as we know it feasable, as we didn't have pest-proof containers back then.

Mixing two subthreads here... is there any evidence that Neanderthals ate cats?

I will mention that I have a techie friend (builder of the world's first electric Ford F-150, which he did over a dozen years ago) who had a cat he raised from a kitten. It's a fascinating picture seeing him wrestling with his fully-grown puma. On the couch in the living room.

Google are physical presence lunatics, but surely they of all people could have managed a talk over the internet

Bringing in meat puppets is also a sign of wealth, you don't get that from a video chat. "I made you spend 30 hours in a box" is part of the value being delivered.

Almost a decade ago a friend of mind was researching this and even flew from Australia to Europe to attend a conference and deliver this paper*. When I suggested that if she couldn't make telepresence work what hope for the rest of us she got quite offended. Apparently important people still get to fly, it's only peons who should be limited to video chat.

* https://research.monash.edu/en/publications/meeting-in-the-global-workplace-air-travel-telepresence-and-the-b

"humans are likely to be dangerous." Or, as I prefer, humans are the deadliest predator in the history of the planet.

Apparently important people still get to fly, it's only peons who should be limited to video chat.

It's hard to chat in small groups or go out for drinks after the talk if it's via video.

FWIW I too interpreted your post (and others passim) as quite clearly meaning what EC thought it did. Perhaps that in itself is evidence that you are a non-verbal thinker... :)

"If people are going to build it or design it, then symbols are required."

Well... if people (plural) are going to design or build it, then symbols make it a whole lot easier. I'm not convinced they're essential, although trying to follow that one to a conclusion quickly gets into stuff like "what counts as a symbol" and suchlike indications that it probably doesn't have one.

However, the ability of two or more entities to communicate is not the same thing as the ability of one entity to think, and the latter is what I think the discussion was about.

Weird evolutionary FPGAs: yes, they end up doing some very strange things. Varying the power dissipation of one area of the chip so as to alter the temperature and so the characteristics of transistors in an adjacent area is another one. Or generating a regular clock from some random agglomeration of the spread of a great big bag of propagation delays. It's entirely possible to end up with a design that only works on the specific device it was generated on, or that only works in the presence of noise from DuQuesne running his utterly unrelated experiment in the lab next door, etc.

This kind of thing isn't about symbols; it's the same deal as the obscure interactions [natural] evolution comes up with that you mention in your next post. It's precisely because "evolutionary design" does not involve intelligent behaviour like planning towards a desired end or adherence to generalised design rules. There is no such thing as "digital electronics"; there's only analogue electronics used in a digital way, and the distinction is created by the discipline of human designers in designing a circuit so that only the digital aspects of its behaviour are significant. But there is no enforcement of that discipline in the configuration logic of the FPGA, so "designing" it by reshuffling random bit patterns until it appears to "work" will explore the whole of the possibility space that the configuration logic allows, including those parts (which are most of it) that an intelligent designer will take care to avoid because using them does not give predictable/consistent/repeatable results.

DNA is of course doing exactly the same thing, but natural selection tests the results more exhaustively and also less rigidly than the electronic setup does. So it will produce results with the same kind of deficiencies, but on occasion instead of rejecting a deficient result it will find a way to pile another bodge on top of it to maintain the right conditions for the deficiency not to matter, or not matter too much. So you end up with a body which has a vital major clock signal produced by some random agglomeration of the spread of a great big bag of propagation delays and it nearly always works, and a function for painting in missing parts of images to cover for your retinas being inside out.

"Apparently important people still get to fly, it's only peons who should be limited to video chat."

Super stuff. I don't want to be an important people and I don't consider getting to fly in an airliner as anything more than another reason for being glad not to be. In fact I'm going to be a total and utter pleb and use only email.

It's hard to chat in small groups or go out for drinks after the talk if it's via video.

You'll note that during covid people tried really hard to make those sort of social events happen via telepresence because they thought it was better than personally facing the threat of death right here and now. Well, in advanced countries, anyway.

So whether it's millions of people should fly around the world every day is a statement of priorities rather than a necessity. We can't afford mass flight until someone fixes climate change, so stuff that relies on it is bad. Or not, for people who DGAF about anything that doesn't happen to them right now.

It's interesting to me that Aotearoa seems to be going through a little bit of "leopards ate my face" politics right now. The biggest city (Auckland) has had a couple of bits of nasty weather, some nearby areas have been hit worse, and a few important people have pointed out the link to climate change. Hopefully that leads to a willingness to take actual action rather than more hopes and prayers.

Unfortunately, us peons is gonna have to fly to come to Glasgow next year, unless you know of an ocean liner, not a cruise ship, going across the Atlantic....

HN* just linked an interesting paper about eye contact in conversation that for some reason made me think of some of us who write here. Not that I've met hardly any of us, but I get the impression there's not lot a of eye contact from some here. {looks vaguely in the direction of whoever's talking}

https://fermatslibrary.com/s/d4a90-eye-contact-marks-the-rise-and-fall-of-shared-attention-in-conversation

Here we test whether eye contact, a common feature of conversation, predicts this coordination by measuring dyadic pupillary synchrony (a corollary of shared attention) during natural conversation. We find that eye contact is positively correlated with synchrony as well as ratings of engagement by conversation partners. However, rather than elicit synchrony, eye contact commences as synchrony peaks and predicts its immediate and subsequent decline until eye contact breaks. This relationship suggests that eye contact signals when shared attention is high. Further- more, we speculate that eye contact may play a corrective role in disrupting shared attention (reducing synchrony ) as needed to facilitate independent contributions to conversation

* HN in Australia normally refers to Hardly Normal Harvey Norman, a chain of appliance/furniture shops owned by a billionaire who likes to whine to the media about his lack of profit and huge tax bills.

That's ancient knowledge - pupils dilate when you're interested in what someone's saying, and one of the reasons that some people wear sunglasses when wheeling and dealing.

Here's an awesome combo of RoboDebt and self-driving: cars that repossess themselves when they're not paid for. Why bother with the legal system or tow trucks when you can repo cars at a company policy level?

One obvious need for legal changes here is what happens to people who get in the way of the policy? From the infant sleeping in the back seat to the idiot standing in the way when the vehicle leaves, who's liable for injuries? And as we saw with the RoboDebt scam in Australia, there's no need for the debt to exist for the policy to kick in, and it will be hard to discover who exactly is responsible if the debt is later found out not to exist. "sorry about that, here's a voucher for a new baby"?

https://www.thedrive.com/news/future-fords-could-repossess-themselves-and-drive-away-if-you-miss-payments

FWIW I too interpreted your post (and others passim) as quite clearly meaning what EC thought it did. Perhaps that in itself is evidence that you are a non-verbal thinker... :)

Thanks. I'll try to be more clear.

It's seems weird to me to be arguing that not all thinking involves symbols (of which words are a subset in my estimation), and to be accused of thinking only in symbols.

Mixing two subthreads here... is there any evidence that Neanderthals ate cats?

I'm not aware of any, although there is evidence that a cat chomped a Neanderthal ( https://www.smithsonianmag.com/smart-news/ancient-carnivores-had-taste-neanderthal-180955390/ )

The study assumed the dialation and tries to correlate it with eye contact.

RE: feline cognition.

One of my favorite puzzles about my current pair of domestic furballs is the following scenario: Around meal time, they show up and solicit petting. Then they beg for food.

In human terms in feels mercenary and transactional: I let you pet me, so feed me. No pay, no play.

The more I work with them, the more I doubt that's what they're thinking. I'm guessing now that they know they're stuck with us feeding them, because I won't let them out to hunt or beg (they're both aging former strays who I've limited to indoor life). So what they may be doing is soliciting grooming to pose as dependent kittens, so that I'll then feed them.

This mode of thought, if true, feels weird to me.

What's weirder to me is that some of the Bronze Age texts I've seen (I remember in 1177 B.C., The Year Civilization Collapsed), the various kings of the ancient Mediterranean wrote letters to each other using the logic of my cats. However trade was actually carried out among them, the letters we have were a formula my cats might recognize: establish notional familial bonds (Oh, my great father in Egypt, you are most mighty and mot generous), then ask for a boon (please send your loyal son more weapons. And gold, if you can spare any).

By the time these letters were written, cats had been domesticated for thousands of years, so I amused myself by imagining that millennia of traders sailing around with cats on their ships taught the humans to think and relate to each other like cats.

Alas, I suspect the real reason this relationship mode seems odd to me is that I've spent too much time trying to pass as a Homo economicus in our modern society. In doing so, I've forgotten what it was like to be a kid and beg for things. Oh well.

David L @ 196:

Apparently important people still get to fly, it's only peons who should be limited to video chat.

It's hard to chat in small groups or go out for drinks after the talk if it's via video.

... or to share conversation over a meal.

Plus, over video I have a very difficult time getting a word in edgewise if I do have a question.

So whether it's millions of people should fly around the world every day is a statement of priorities rather than a necessity.

While this sub thread started out talking about a Edinburgh to San Jose trip for an in person presentation, there are a LOT of local things which are still a thing.

There were multiple tech groups (DevOps, PowerShell, Lunix, etc...) that were very active around here pro-Covid. All of them mostly fell apart or shrunk way down during Covid and video conferencing. For the reasons I mentioned. Conversations over the pizza buffet were as useful many times as the main meeting topics.

Especially the job offerings and seeking conversations.

Video meetings can be and many times are very useful. But in so many ways are just not the same.

Re: 'That's probably not important to an AI, but evolution is full of very indirect interactions, so we can't REALLY be sure'

Okay, I see you've taken the reductionist road here - chemistry*. But that still doesn't answer the variety of different inputs that humans can and do use to interface with their environment and each other vs. what AIs use for inputs and interfaces.

*Actually, you can take this a step further down into physics land, i.e., sound, light, weight/pressure, balance/direction, etc.

Anyways, it'll be interesting to see at what point AIs come up with specialized segments of their data processing innards --that is, see whether AIs follow the same evolutionary paths as organic beings.

I looked for articles on the evolution of the nervous system. Not much online but this book looks interesting - unfortunately it's not available at my local library.

Dawn of the Neuron: The Early Struggles to Trace the Origin of Nervous Systems (Michel Anctil)

The concept of "dictionary" something so simple that it was obvious after you explain it to someone. But. But first someone to come up with it and then go through a couple dozen drafts for how entries ought be laid out. It is a critical tool in leveraging the fullest value of language. Most folk rarely need much of a vocabulary, just about 5000 commonly used -- most heavily used -- words in living their typical days. Only if they need to converse with a specialist in some aspect of our civilization do they get dragged out of their comfort zone. And not everyone is ready to locate an appropriately helpful dictionary. They try to muddle along, never mind understanding what a car mechanic or a cancer specialist doctor or a laptop technical is trying to explain to them. Just one more instance of how most humans really are not good at tools, not ready to leverage their available technology.

Technology is a matter of being clever with tools and constructing useful (and usable) tools. The revised definition of "human intelligence" is not simply 'tool maker' but 'making tools with which to make better tools'. Whether communications based upon words between nominally sapient individuals or data crunched by marginally sentient artificial intellects, it a matter of the tools available.

Given how different parts of an infant's brain develop at differing rates but from their first breath they are seeking to communicate, plus frustrated parents and thus you get BSL. An extraordinary instance of a deliberately improved tool kit: BSL (https://babysignlanguage.com/about-us/).

Arguments about chatGPT (and other supposedly operational AIs) rarely detail much about what is in their 'tool kits'.

201 - I'm not definitely aware of a shipping line that actually still has a proper ocean liner in service. Worldcon time is about the one time of year when a cruise ship can cross the North Atlantic to schedule, without risking being bent in the process.

204 - Repossession might work a bit differently; Repo Man descends on data centre for service of $vehicle with court order stating reg and chassis numbers, at which point data centre instructs the vehicle to, when unoccupied, lock itself and report to the car pound, where it will promptly be rechecked for nominally domesticated life (and cats) before being handed to the organisation posessing the court order.

212 - Not sure I entirely agree here. I know a bit of chemistry, maths and physics, but only a little of biology and medicine, and can still usually keep up with nurses and surgeons explaining a procedure to me.

213:

if you really crave ab opportunity for snarking upon the youngest generation... “tech shame” as per article... which given my experience from 1985 thru 2010 is an understatement of people's cluelessness and what amounts to 'learned helplessness' due to being spoonfed for too many years of higher education without enough hands on apprenticeship... such as MBAs unable to load printers with paper... https://www.theguardian.com/technology/2023/feb/27/gen-z-tech-shame-office-technology-printers

the brutally sad thing is most adults in US have a tough time dealing with the 'horrid trifecta' of sudden medical crisis (emotional) + insurance paperwork (bureaucratic) + have little clue what various organs do (medicine)... each having its own vocabulary and takes time to learn which is not available in midst of crisis

Repo Man descends

I think the idea is to remove repo man from the system altogether, and use self-driving to mean self-repossessing. And no doubt self-stealing when other criminals get involved.

A lot of large corporations who nominally sell things make most of their money through their financial side. GE is probably the best known, where various flavours of "GE Finance" will help you buy almost anything if you have a credit rating they approve of. But if in this example Ford can reduce the cost of lending people money to "buy" their cars that's money in the bank for them. Self-driving just lifts it from immobilising their vehicle with all the attendant risk of damage and legal trouble, to self-repossessing which is likely to be largely unregulated.

"buy" because of course with a computer on wheels you only kind of technically own a limited title to some physical parts of the vehicle, subject to legally binding conditions imposed by the manufacturer. You can't, for example, modify the software without approval or install non-approved spares for some parts.

They try to muddle along, never mind understanding what a car mechanic or a cancer specialist doctor or a laptop technical is trying to explain to them. Just one more instance of how most humans really are not good at tools, not ready to leverage their available technology.

There's more to it than just "not using available technology".

Technical fields tend to generate their own fractal levels of jargon. Sometimes that means made up words unique to that field. Sometimes it's an abbreviation or acronym that gets turned into a word. And sometimes words just get redefined to mean other things. As a physicist-turned-neuroscientist-turned-programmer, I can think of about 5 different meanings for "coherence".

The same is true when working in other languages, or even just different dialects of the same language. For a trivial example, consider that Americans tend to use "could care less" when the concept they are trying to convey means the exact opposite of what the words they're saying actually mean, i.e. that there is no lower level at which they could care: that they could not care less. Use that phrase to another American and no confusion is engendered. Use it to a British-English speaker who's never encountered it before, and there will probably be a double-take. Use it to a non-native English speaker who has to work through a sentence word by word, and they will take away the exact opposite message to that which you intended.

I would also note that communication is a two-way street: if you're having a conversation with someone else and the other person clearly doesn't understand, then you're failing to communicate with them just as much as they are failing to understand you. A good communicator will try and work out which bits are causing the lack of understanding - and try to find a way to re-express that using concepts that are more likely to be common between both (or all) participants.

Ring LASERS for Time Travel! - well, maybe, maybe not, but it's fun ...
Unless & until it works - & THEN we get the unpleasant answer to Fermi's Paradox.

Howartd NYC
Tell me about it ....
Having done a little workshop practice at school & then working in research labs, where, a lot of the time, you built your own equpment - or at least assembled it out of scrounged bits ...
No-one has the faintest effing clue, do they? Explains why & how plumbers & (domestic) electricians can make piles of moolah, though.

Unfortunately, us peons is gonna have to fly to come to Glasgow next year, unless you know of an ocean liner, not a cruise ship, going across the Atlantic....

Lots! Unfortunately the Queen Mary 2 is the only one still taking passengers; the cargo liner booking services all have "no passengers due to covid, try us again in 2024?" on their websites.

Then Charlie insisted that the only way Homo sap. sap.'s gene parents could have bred was by rape.

That's not my position. My position was that describing inter-hominin matings as "friendly" was almost certainly wrong some of the time. Quoll at #168 more or less sums it up. Rape is unfortunately regrettably common among H. Sap., although the figure I've seen doing the rounds is that less than 3% of the male population are rapists: I see no reason to believe the population incidence of rapists was lower in prehistoric or pre-Sapiens times than it is today, and glibly dismissing all sexual interactions as "friendly" looked like rape-washing to me.

(I suggest we drop this topic.)

I don't see any reason why the failure of crypto proves that AI will be a flop as well.

That's not my argument.

I'm arguing that the shady make-money-fast grifters who were pushing cryptocurrencies have moved on to promoting AI, in an attempt to get rich quick (at the expense of more gullible investors).

Also, what we're seeing isn't "AI" in any meaningful sense, it's just the use of deep neural networks for machine learning on a scale that was hitherto unimaginably vast: Moore's Law finally caught up with where we needed to go in order to get stuff like working machine vision, plausible language models for grammatical text generation and translation, protein folding, Hansch analysis/QSAR, and so on.

It's nothing like AGI and isn't a plausible route to AGI (even assuming AGI is possible).

I know Google are physical presence lunatics, but surely they of all people could have managed a talk over the internet?! Talk about self-defeating.

This was well over a decade ago. No Zoom, no Facetime, I'm not sure what video conferencing tech Google had but I didn't have it and nobody was suggesting it. Oh, and home broadband was a lot slower.

When I first got home broadband in 1998 I was overjoyed to have a 256kbps/64kbps (down/up) cable connection with 50:1 contention ratio instead of my 56kbps model. Today I'm kind of "meh" about my 300mbps/80mbps FTTP line with no contention ratio and no throttling/metering either ('cause I'm paying extra for a business grade package because, uh, I'm a home business). Oh, and automatic fail-over to a 4G/LTE backup if the line goes down. We've come a really long way in the past 25 years ...

Also, what we're seeing isn't "AI" in any meaningful sense, it's just the use of deep neural networks for machine learning on a scale that was hitherto unimaginably vast[.]

Yeah. I did a couple of graduate-level neural network courses about 25 years ago, and I see the basics there in the current ones, just the scale is quite much larger than what we could run at the time as our test networks.

These systems have their uses, but I think it's easier to get funding for those fancy Markov chain chat engines and image generators than for real expert systems. Especially as you'd need some kind of domain knowledge for the expert systems, but with the hype ones you can use the GIGO principle, without consideration is your output correct or not.

I just hope the hype dies down sooner than later, along with the drivel from them.

If one of those proposed devices were practical enough to see if Einstein's formula (as distinct from theory) applied up to the point of singularity, it would be an incredible achievement. My prediction is that it won't, and would show that it doesn't, anyway,

I doubt that we shall see one before fusion power becomes the norm ....

Exactly, though it goes back to at least the 1960s. People knew what could be done, in theory, and there were some practical applications even at that date. Yes, of course, we have learnt that a great many of that era's assumptions were wrong, and people have thought of new uses, but most of it is really just an application of simple principles on a vast scale.

I know how to make a sailing boat go; I'd suggest the nautical equivalent of hiring a minibus, but I'm on the wrong side to start with...

Hmm. I tend to spend a good deal of the time in conversation with my eyes closed completely. It makes it easier to allocate enough CPU to the conversion between thought and language.

However trade was actually carried out among them, the letters we have were a formula my cats might recognize

Huh. That sort of negotiation rings a bell from Graeber on Debt? It's not a transactional business exchange (goods for money) in the modern sense we recognize, but being able to graciously give the requested stuff burnishes the king's prestige (and in turn, makes it more urgent for others to pay tribute to him).

Conversations over the pizza buffet were as useful many times as the main meeting topics.

Exactly.

Since COVID I've partipated in SF conventions mainly over Zoom (with a side-order of Discord, and a handful of hellish alternative virtual conferencing apps which never work well because Zoom and Discord are best-of-breed for a reason). Which is inconvenient, because you can never get the serendipitous interplay of meeting strangers face-to-face via Zoom, let alone of walking into a conference room halfway through a presentation or panel that didn't look that interesting in the program only to discover that it's actually vitally important to your interests.

Virtual meetings are single-focus/single-track. Whereas an in-person conference offers all sorts of unexpected opportunities.

plus frustrated parents and thus you get BSL.

Presumably you're not talking about British Sign Language, right?

(It's best to expand all TLAs -- three letter acronyms -- on first use, because the namespace is full of collisions.)

213 Para 1 - I'm not surprised that MBAs can't load a printer. Specialist consultant doctors have difficulty understanding that, if you want a printout in your department in hospital_1 you have to change the printer assignment from hospital_2 to hospital_1!

Para 2 - Yeah, I'm still learning dialysis stuff after 2 years.

215 Para 3 - Actually yes you can modify software for a vehicle, at least in Europe.

I think the idea is to remove repo man from the system altogether, and use self-driving to mean self-repossessing. And no doubt self-stealing when other criminals get involved.

BRB, just had an idea for a horrific near-future short story in which this tech is used for administrative punishment of driving infractions (speeding, parking tickets, not keeping up with your car loan repayments) ... and the court can order a driver to be imprisoned in their vehicle. Which periodically drives to a charge point, or through a drive-by prison canteen window (if you need to take a shit the meal trays double as flat-pack biodegradable commodes), or just sits in the sun, slowly cooking you in public. Car as mobile solitary confinement prison cell, in other words, except it's in full public view all the time and the central locking won't let you out before your time's up. (If you smash a window to get out, that's a jailbreak and will earn you real prison time.)

With work-from-home you can of course be employed in your prison cell, tagging metadata for a deep learning model or answering customer helplines or something.

Of course, if the wrong person is driving the car when it goes into punitive lockdown, they might end up serving a month long jail term instead of the car's owner. And if the car's owner has a heart attack and dies at home in the meantime they won't be able to answer their exit processing questionnaire ("are you suitably sorry now you've served your time?") and provide confirmatory biometrics in order to be allowed out of automotive jail ...

The Lord's Prayer follows the same pattern. So do an awful lot of requests from historical periods I'm more familiar with. "Prince Prot of Berrovia eventually succeeded in subduing the rebellious Strobererian province after requesting military assistance from the Gobro of Gwed", say the basic texts. When you find a more detailed one which quotes Prince Prot's actual request, it turns out to be half a page of extravagantly florid verbosity felching the Gobro cleaner than a salt-cellar, followed by asking for the moon on a stick. The start of the pattern's decline in popularity seems to broadly go along with the rise in the popularity of chopping kings' heads off if you didn't like them and choosing your own replacement or doing without kings altogether, as the contempt of one's fellows for such extravagant expressions of subservience grew to overwhelm the importance of the recipient's delight in them. However, it still remains with us, albeit in a highly attenuated form, hence the entertainment value of Neil's letter to his bank manager^* from its scrambled approach to modern conventions.

(^*"Darling fascist bully boy. Give me some more money, you bastard. May the seed of your loin be fruitful in the belly of your woman.")

It's been done... without the computers. Accidentally fart in public and you get to drive continuously on a network of dedicated highways, stopping only for fuel... with your entire family. So you get the nightmare holiday trip with a car full of fractious kids, dragged out into months or years of torment.

I will note that ocean liners are famously neither carbon-neutral nor low-energy: you'll burn far less fuel crossing the Atlantic on an airliner (and it's cheaper and faster).

People knew what could be done, in theory, and there were some practical applications even at that date.

Yes and no.

Marvin Minsky set neural networks back decades when in the late 1960s he came up with a clear proof (not disproven since, either!) that a perceptron-based network was computationally too weak to learn an XOR system and therefore wasn't capable of general-purpose computing. Perceptrons are single-layer neurons but Papert and Minsky incorrectly generalized this result to multi-layer neural networks. And two such eminent computer scientists come up with a good proof and then stacking a not-obviously-false conjecture on top of it killed interest in NNs until the late 1980s.

Of course, single layer perceptron networks are not the type of neural networks that are in widespread use today -- post-1980s NNs all have one or more additional hidden layers.

Video meetings can be and many times are very useful. But in so many ways are just not the same.

The Ontario Association of Physics Teachers is holding the first full in-person conference since Covid, for precisely those reasons. OTOH, they're also keeping up the Virtual Physics Hours that were started as a replacement for in-person meetings, because they are also useful.

The virtual meetings are a lot cheaper to attend, which is important when the costs of attending usually come out of the teacher's pocket. They also allow people to easily attend from far away — despite the OAPT being founded for Ontario, the last virtual event had people from thousands of kilometres away join in.

TLDR: in-person and virtual conferences/meetings serve different needs. Just as in-person and virtual classes.

http://newsletter.oapt.ca/files/2023-conference-announcement.html

http://newsletter.oapt.ca/files/physics-hour-march-2023.html

Kill switches are already used to disable cars who's owners are late on payments, which has led to some interesting legal issues. Borrowers have disabled the vehicle but not repossessed it, while trying to collect the full amount of the loan from someone who now has a non-functional hunk of metal rather than an asset that can be used or sold.

It seems rather like your mortgage lender ripping your roof off (rendering your house uninhabitable) but still insisting that you repay the full amount that you borrowed.

https://www.nbcchicago.com/news/local/car-dealer-lender-deactivated-cars-with-kill-switch-after-borrowers-fell-behind-on-payments-lawsuit/2881440/

A legal opinion:

https://www.youtube.com/watch?v=kYuzaM2FcWU

It's not a transactional business exchange (goods for money) in the modern sense we recognize, but being able to graciously give the requested stuff burnishes the king's prestige (and in turn, makes it more urgent for others to pay tribute to him).

Sounds a bit like potlatch over a wider scale.

A few years ago I backed an interesting-looking game on Kickstarter, designed by indigenous gamers to teach their culture.

https://www.kickstarter.com/projects/473205107/potlatch-a-card-game-about-coast-salish-economics

The 'buy now' link on Kickstarter is broken, but here's one that works if you want to try it for yourself. I found it a fun game:

https://www.drivethrucards.com/product/182080/Potlatch-A-Game-About-Economics

I wasn't talking about the theoreticians, who were usually away with the fairies (*), but the practioners - including a great many people in unexpected fields (e.g. psychology, linguistics and social sciences). Yes, I do mean that Psychometrica was a more useful reference than any 'computer science' journal!

Route and schedule optimisation etc. date from that period, as do adaptive help/search systems. Except for a few high-value cases (optimising grain shipments to Israel is one I remember), it wasn't until the 1970s that they started to be used much, because of computer constraints, but we and a few others started using them then.

(*) That aspect got worse in the 1980s, too.

Of course, if the wrong person is driving the car

Ah, but access to the car is only possible in the first place when the face recognition system in the vehicle matches the potential driver to the biometrics from their driving licence and confirms their identity as an authorised and correctly insured operator of the car.

Beware if your identical twin sibling seems very insistent on you borrowing their car...

It's not just repayments - it is starting to be applied to to cars that have been paid for in full! Several vendors have contracts that say important features require regular payments if they are to continue working. They haven't yet gone so far as applying that to essential features, as far as I know.

Yeah. If your mode of transport takes so long that you have to carry along beds and showers and eateries, the per-person load will increase. This applies to long distance trains too - multi-day ones, that is where you have dining cars and showers and cabins.

It's considered that the only ocean liner today is the Queen Mary 2. Otherwise, if you want to go transatlantic, you're probably looking at one of the cruise liners, and they stop everywhere they can because that's their point: a five start hotel with (where possible) a new place outside every morning. I've gone that way (Bergen to Montreal) and it's lovely, but it's not the way to do it cheaply.

Maybe a bunch of fen could get together and charter a plane? No idea what it would cost or whether it's a good idea, just thought I'd throw it out there.

Sounds a bit like potlatch over a wider scale.

I got the game, thanks!

In a PNW-produced textbook on PNW indigenous design, I saw Potlatching described as more like capitalism than it is usually presented. While this is apparently a standard text for teaching aspiring tribal artists how PNW design works (I bought it in Victoria years ago at a Native art gallery), it's the only place I've seen this description, so take it for what it's worth.

Potlatching is normally described as competitive partying, with feasting, massive gift giveaways, entertainment, rituals, and important people giving each other important things. If the recipients of largesse failed to reciprocate with at least an equivalent party later on, they lost status, while they gained status if they threw an even more lavish party in a few years.

Realize that what we're talking about here are basically chalcolithic chiefdoms. They worked copper, were socially stratified, kept slaves from as far away as California, had agriculture (this wasn't recognized until recently, but the women cultivated a variety of bulbs), and had mariculture (extensive constructed clam beds, giant-scale fish traps, etc.). So not simple fisherfolk.

Anyway, the art book represented potlatching as (in part). A chief would give an ally a resource, like a berry patch or whatever, along with baskets of berries from the patch. The terms of the deal were that the ally was then supposed to use and manage the berry patch so that at the next potlatch, they could give back the berry patch, plus at least as many berries as they'd been gifted. This was phrased as a venture capital gambit in the book: the receiver of the gift was supposed to manage the gift not just to support themselves, but to be able to give back a profit to the one who gave them the gift originally. Paying back an investment with interest, more or less. Of course people who couldn't pay back the original stake were shamed as a result.

The point of doing this during a party was to have dozens to hundreds of people would witness both the original deal and the payoff at its end. The point of all the other things going on was to keep the witnesses hanging around so that they would witness all the business deals (aka gift giving) going down between the medicine shows, entertainment, and ritual. After all, without writing, only peoples' memories kept track of who owed who what and when. Having lots of witnesses to business deals was therefore important..

I think many medieval rulers might recognize this format? Whether the tribes who potlatch actually agree with this is something I as an outsider can't say. But I think it's worth thinking about maybe.

Thus why cargo liners: the per-passenger load is absolutely swamped by the actual day job, and the passenger numbers are miniscule anyway - there tends to be less room for passengers than for crew. I just enjoy that there are whole travel agencies out there for booking berths on cargo ships, no matter how impractical travelling that way would actually be. :-)

(Slow, no-frills, and expensive (>€100 per day); just what one wants in a transportation option.)

Charlie, your most recent post got cut off: "Of course, perceptron networks are"...

So, the obvious next question: did cat domesticate Neanderthals, or not? Are what seem to be pets/companions among the Neanderthals?

5000? I've read, years back, that the average American's normal vocabulary was about 500 words, not 5000.

And I've always enjoyed pointing out that I also read that Koko the gorilla had 550 words in ASL (sign language).

Repossession - I can think of so many things that will go wrong with this.
1. Parked in garage with door shut.
2. Parked on long gravel driveway, and gate.
3. Has to drive on three-lane street (like out front of my house, with two lanes for parking).
4. All the rest of the reasons self-driving vehicles can't/shouldn't drive anywhere but on limited-access highways, including, say, child runs out from between parked cars.)

Ellen, my partner, like my last ex, wants Iceland Air, with a stop in, well, er, Iceland.

I'm not entirely sure why I'm reminded of the tributary system formalized under the Ming dynasty, where embassies from foreign powers who formally submitted and gave tribute to the Chinese emperor were acknowledged as rulers of their territories under Chinese authority and given gifts and permits to trade with China; such gifts and permits were worth so much the Hongwu Emperor restricted some countries in how often they could give him tribute.

Here's a very interesting article analysing American foreign policy through the lens of this tributary system...

5000? I've read, years back, that the average American's normal vocabulary was about 500 words, not 5000.

Ogden identified about 850 words as being sufficient for most purposes.

http://ogden.basic-english.org

For a while I tried to use it when writing handouts, because I had a lot of ESL students and I wanted things to be easier for them to understand. It is hard to write for someone used to using a larger vocabulary and more complex grammar. Eventually I stopped bothering with Basic English and just concentrated on writing simply and directly.

Ellen, my partner, like my last ex, wants Iceland Air, with a stop in, well, er, Iceland.

I flew Iceland Air to Iceland. It was a wonderful trip; one that I want to repeat. Iceland is amazing — so many different things to see crammed into such a tiny island.

Here's a very interesting article analysing American foreign policy through the lens of this tributary system...

Sorry, but there's no "here" there, just a URL not found on server message…

Several vendors have contracts that say important features require regular payments if they are to continue working.

This has been true for 20 years in the US. GM's OnStar.

And more and more thing require a cell connection. Either they bundle it in up front (Tesla isn't cheap for many reasons) or you pay a monthly or annual fee after the initial period runs out.

5000? I've read, years back, that the average American's normal vocabulary was about 500 words, not 5000.

Are you saying common use every day (which Howard was meaning) or words they can understand the meaning of and use if needed.

Most people in the US can deal with at least 25K words. College educated folks go to 70K to 100K and there are a non trivial number of people who can deal with 125K or more.

And the followup to what you wrote is the same as a video we were required to watch at one job I had: Eschew Obfuscation.

Iceland Air, with a stop in, well, er, Iceland.

Due to its cheap fares Iceland Air gets lots press. But folks who fly it basically say it is the international version of RyanAir (Europe) or Spirit (US).

By having nearly all flights stop in Iceland they get to use smaller planes than those flying all the way across the Atlantic.

Yeah, my ex, stepson, and I were there for about a day and a half in '14 on the way to Worldcon. I took the three-whatevers tour - they went glacier scrambling (that was before I had both knees partially replaced....)

I should complain - no trolls, elves, svart-elves...

If you are adventurous and have the time there are many hundreds, even thousands, of sailboats crossing the Atlantic from West to East every Spring. They also go back the other way in the Fall (usually after hurricane season is over). When I were a younger fool I did precisely that.

Ah damn, it's added on to the end of an antipope.org url; I must have screwed up my html. It's https://academic.oup.com/cjip/article/6/1/1/457227

I hear Icelandair is a good way to go trans-Atlantic if you want to visit Iceland en route. However they fly 737s, so you're looking at spending 4+ hours aboard a narrow-body with economy class seats only (no premium economy or business class let alone first class) on both sectors. I guess it depends on how short/skinny/flexible you are? Because if you're tall/fat/creaky, it may a less-than-great idea.

Indeed: actualise eschewing otiose prolixity for the achievement of perspicacity

Actually, IIRC, they offered three choices of leg room (the larger, of course, more expensive).

Re: 'Perceptron'

Okay, so one of the built-in issues I have with AI inputs and performance has a name. Thanks!

Per Wikipedia: 'Single-layer perceptrons are only capable of learning linearly separable patterns.[6] For a classification task with some step activation function, a single node will have a single line dividing the data points forming the patterns.'

About early hominids in Europe and why only one hominid group survived:

I think the likeliest comparable scenario is the European 'conquest' of North America - germs not weapons.

Another additional factor could be subtle genetic differences -- Not sure when or where Rh Neg factors originated but if one hominid group had them and the other didn't, there'd be a serious problem with interbreeding. The first pregnancy would progress to term and that's it: only one child in a high infant mortality rate environment. Such reduced odds of successfully having a child would likely become a major barrier to interbreeding.

So, the obvious next question: did cat domesticate Neanderthals, or not? Are what seem to be pets/companions among the Neanderthals?

Poor kitteh keep up with mammoth hunting hooman? In snow?

No evidence for any hominid domesticating or even taming any species while the Neanderthals were extant, so far as I know.

I will say though that neotenic, symbiophilic Neanderthals and/or Denisovans are high on my list of alt.history ideas worth exploring.

Here's a very interesting article analysing American foreign policy through the lens of this tributary system...

Thanks for the link and the fix. I think it's a reasonable critique. I think we have to recognize that, bullshitting aside, the US is the biggest military power probably the world will ever see, unless we get really good at fusion power, batteries, and not making the planet unlivable. That's not rah rah for us, it's just that we peaked at the right time relative to oil and other resource extraction, and decided to waste most of it on weapons and other consumer goods.

I'm sure that will be real comforting for people who spend the next ten thousand years (or more!) dealing with all the garbage we're creating in this role. And so it goes.

Video meetings can be and many times are very useful. But in so many ways are just not the same.

This can be particularly painful for participants in wildly different timezones to where the phyical meeting is taking place. This hit a number of conferences when lockdowns arrived in 2020. It also hit audio/video media, like podcasts, where guests could be (and frequently were) on different continents.

For comedy, video conferencering presented other problems, like delayed laughter or no audio from the audience at all.

OTOH Zoom got what one podcaster described as a decade's worth of "development" (I think that was the word used) in the first month of lockdowns.

Due to its cheap fares Iceland Air gets lots press. But folks who fly it basically say it is the international version of RyanAir (Europe) or Spirit (US).

I found the Toronto to Reykjavik price from Iceland Air about the same as the Air Canada price. And the service was about the same, with the advantage that I could buy a SIM on the plane rather than having to queue at the airport.

Didn't strike me as a budget airline, really. First checked bag free, in-flight meal included, etc. Nothing at all like Air Transat, which is what I think of as a budget airline.

"I wasn't talking about the theoreticians, who were usually away with the fairies (*), but the practioners - including a great many people in unexpected fields (e.g. psychology, linguistics and social sciences)."

I think the big turning point back from Minsky was Rumelhart and McClennand, mid 80s. Using the back-propagation techniqus for multi-layer neural nets from Werbos (late 70s).

They used back propagation to teach a neural network basic phonetic rules for reading (for simple words containing a subset of letter). Actually: to convert simple words that use only a subset of english letters, into a subset of the international phonetic alphabet (IPA) that reflects how they are said. They showed that the neural-net then over-generalised (by trying to use simple strategies, but actually that's wrong for many English words because our real phonetic rules for reading are insanely complicated).

5-year olds learning to read also over-generalise by using simple rules. So their neural net would make the same sorts of mistakes reading that a 5-year-old makes.

Ergo, obviously neural nets learning to read are like kids learning to read! Using things like neurons! Neural nets model what the brain does! Quick, let's get more research funding!

The huge circus was that they did talks and demos at huge psych and linguistic and cog sci conferences with the neural net "reading", where they hooked up the output to a voder/vocoder that "read" the words in the voice of a small child. Making the same types of mistakes that are made by a small child. It was a huge spectacle, everyone aahed and oohed. Immense interest in neural nets resulted in the late 80s and 90s in psychology, linguistics and cognitive sci programmes.

One can get quite cynical about what drives fashions in science and academia.

This can be particularly painful for participants in wildly different timezones to where the physical meeting is taking place.

The conduct of a physical meeting is different to a virtual meeting. Trying to mix them doesn't really work, unless the 'meeting' is really a presentation with selected questions from the audience.

In my experience hybrid classes (mixed online and in-person students) have been a dismal failure. Likewise I've never heard of a hybrid meeting that really worked.

In an organization where personal connections are important, being the remote party is basically being relegated to powerlessness — as many people who've been 'decentralized' can attest to.

However, neural networks are only one technique of many and, despite the hype, do NOT enable anything that can't be done in more 'conventional' ways. The practitioners in the 1960s and 1970s used less memory-hungry methods, for obvious reasons. I worked in the same department as one of the leading researchers into computational linguistics, and the fact that English can't even be parsed (or pronounced!) without considering its semantic context was known very early.

One can, indeed, get very cynical about academic fads ....

There we go - I'm right. Cats domesticated sap.sap, and we could srore food longer, and we outcompeted N.

I'll have to disagree, but I'm looking at what may be smaller zoom meetings. BSFS (maybe 40+ participants)and WSFA (two dozen, maybe) had virtual, then hybrid meetings, and still do. Works. The course I took from Odyssey Online - but that had a dozen and the instructor, even though one was from Greece.

It's difficult to explain things like the rhesus system using simple genetics and evolution, so any simple theory involving it is likely to be wrong.

Way back when, I couldn't understand why factor VIII haemophilia is so common, until I discovered that it's a very common spontaneous mutation. But the rhesus system is not subject to such a simple explanation. Nor can it be explained by recent population mixing, because of evidence against that. Maybe someone has a clue, but I have never met them, if so.

There we go - I'm right. Cats domesticated sap.sap, and we could srore food longer, and we outcompeted N.

Well, this gets super-relativistic. The oldest evidence for cat domestication is around 9500 years ago, while the latest known Neanderthal fossil is around 37,000 years old.

So your theory requires cats to manipulate both time and space. However, it's not impossible, because it's a fact that they can do just this.

That's just so obviously wrong. Cats were on the 'Net before 2001.... https://www.gloveboxoptions.com/shopping/impact-printer-products-ascii-art#ascii-art-product-AART116

Also, what we're seeing isn't "AI" in any meaningful sense, it's just the use of deep neural networks for machine learning on a scale that was hitherto unimaginably vast: Moore's Law finally caught up with where we needed to go in order to get stuff like working machine vision, plausible language models for grammatical text generation and translation, protein folding, Hansch analysis/QSAR, and so on.

It's nothing like AGI and isn't a plausible route to AGI (even assuming AGI is possible).

Don't be so quick to dismiss this just because we kinda sorta understand it. First, intelligence has been defined from ~1950 to ~2006 as "everything a human can do a computer can't", and AI was considered non-existent. Then with DNNs everything got called AI suddendly. It's not a technical term at this point, it's a marketing term.

As for those new models, remember they're new. They don't reason (that we know of), they're though an interesting take on what "intuition" may be. And given that half of reasoning looks like after-the-fact justification of something you've decided intuitively, we're possibly more than halfway there.

Also, the usual question is do we want AGI? We most definitely don't want self-motivation, and there's the usual "planes don't flap wings". How much more is needed to make of such systems really useful tools? I don't pretend to know where it's going to go, and I'm fairly certain nobody does, really.

There sure is going to be scams, there sure is to be a lot of wasted money, but at least compared to the crypto crap these systems actually do something we haven't seen before. How valuable it actually is is an open question.

These systems have their uses, but I think it's easier to get funding for those fancy Markov chain chat engines and image generators than for real expert systems. Especially as you'd need some kind of domain knowledge for the expert systems, but with the hype ones you can use the GIGO principle, without consideration is your output correct or not.

That's also because expert systems tend to be kinda crap. Because human experts half the time are unable to say why they give a particular answer, so they build a bullshit after-the-fact reasoning to justify it. While the reality is that they've seen lots, and while they can't remember the precise experience(s) that bring them to their intuitive conclusion, nethertheless they're most often correct than wrong because we're damn good at pattern matching, recall and interpolation.

And it looks like those new systems are also damn good at pattern mathing, recall and interpolation. They're "just" missing introspection and reasoning, which prevents them to see when they're spouting complete crap.

Also, what makes GPT-3/ChatGPT so much more efficient than GPT-2 is the work that has been done on the quality of the data. So GIGO... not really anymore, that was the prevuois generation.

Also, the usual question is do we want AGI?

I'm about 13-14 years ahead of you.

(I strongly suggest you read "Rule 34" before continuing on this path ...)

Imagine a Q&A following a paper presented at a science conference. It seems some conferences in 2020 simply switched to Zoom, causing some partipants to struggle.

Perhaps a different model might work better, like YouTube/Vimeo videos for each presentation and the Q&A in the comments. That's very different from a video conference, of course.

I lost track of the domain, so I can't comment how the conference world adapted. I can only say that the comedy stand-up and podcasting world eventually adapted rather well, based on what I've heard from comedians on their podcasts. E.g. more than one podcast where particulants were on three different continents. I don't know how many problems simply didn't make it to the final edit, but the resulting episodes sound smooth.

The hardest part may be just getting everyone online at the same hour. Waking up at 5am to do a comedy show doesn't sound trivial to me. Now think about how late someone else must stay awake in their timezone. Two or three different continents, one comedy gig! Everyone interacting and being funny? Yes, I find that impressive.

Likewise I've never heard of a hybrid meeting that really worked.

A club I'm in was meeting monthly in person. Then Covid. We switched to weekly Zoom. (Stir crazy people craved the contact.) Now we're back to monthly in person but have kept the weekend Zoom for the non in person weeks. And for in person we Zoom it.

But the in person Zoom is really a one way feed with maybe a few questions. No real conversations like the Zoom only meetings.

Charlie @ 228
Ms "Teams"???

Likewise I've never heard of a hybrid meeting that really worked.

I've been in a bunch of them. The group I'm with has a number of people who actually care about climate change, as well as a bunch of people who are idiots about it and can't stand not traveling long distances. Our mixed meetings use Zoom Owls ( https://owllabs.com/products/meeting-owl-3 ) and work just fine.

I'm betting you have a strong or dedicated moderator / setup person / they who make it all work. Our small group is more of an everyone steps back when work is requested. :)

Whitroth --

I agree that automated repossession is an evil idea out of dystopian cyberpunk, but your list of "thing that will go wrong" shows you are behind the times.

1. Parked in garage with door shut.

No self-driving car will move in such situation. My Tesla certainly will not -- and will beep loudly to alert humans that it's being given impossible instruction.

2. Parked on long gravel driveway, and gate.

Same thing. It will get to the gate and stop.

3. Has to drive on three-lane street (like out front of my house, with two lanes for parking).

OK, here maybe you have a point. But I doubt it.

4. All the rest of the reasons self-driving vehicles can't/shouldn't drive anywhere but on limited-access highways, including, say, child runs out from between parked cars.)

My Tesla is BETTER at spotting children (also adults) on the road than I am. Several times by now it slowed down for no apparent reason, and THEN I saw the pedestrian. And I've been driving for 35 years, and never hit anyone.

A bill has been introduced in the Montana legislature to ban anyone who has received a Covid-19 vaccination from donating blood.

House Bill 645 would ban individuals who received the COVID-19 vaccine from donating blood, making it a misdemeanor with a $500 fine to donate or accept blood from vaccinated donors. The bill would also ban people who have had a diagnosis of “Long COVID,” medically defined as “postacute sequelae of SARS-CoV-2 due to chronic 27 SARS-CoV-2 viral infection.”

In addition to creating a severe shortage of blood in the state, opponents said there’s no way to test blood for both Long COVID or the vaccines.

As someone pointed out in the comments on another blog, "The states are the Lavatories of democracy."

Climate change and the Agave plant.

I just hope Juniper berries aren't next!

I'll also point out that car companies have the same (sub)stellar record for cyber-security as most internet-of-thingamajiggers do: in other words, it's crap, and people regularly publish papers naming and shaming the problem companies: https://samcurry.net/web-hackers-vs-the-auto-industry/

Anyway, being able to remotely brick or repo a car while someone's in it could basically be used for anything from kidnapping to assault, murder, or act of (info)war. Couple that with serious problems with the rollout of true self-driving cars, and the idea seems to be a menu of disasters looking for fiascos to chest-burst their way out of.

Well, I did say it is a concept out of a dystopian cyberpunk, didn't I? Just not for the reasons Whitroth listed.

Yes, you might be kidnapped in your own car. But the child running out onto the street in front of it will likely be safe.

YouTube/Vimeo videos for each presentation and the Q&A in the comments

That's what I'm used to. It makes moderating the audience much easier and the answers part usually flows better (requires a non-talking moderator or two speakers who alternate). Honestly, the local government zoom meetings worked so much better. Not just the chair being able to mute council members who were naughty, but the cliche "this is more of a statement" turning into a comment from the chair "J Random made a statement supporting {position}". Much quicker :)

I think a lot of it depends on whether you're there for the experience or there to get things done. The former is much harder to do over multi-person video, at best you get contact detail for people you like the sound of. A friend's 40th birthday was both awesome and terrible that way, they ended up with google, microsoft and zoom video sessions going and over 100 people online. Fine for the guy running it on a 300/100 fibre connection, but most of the rest only had one window open and just watched. Not really a social event so much as a TV show.

But the "meeting to decide"... oh boy. Most presenters seemed to adapt quickly and meetings got shorter because of less pissing around. Most people quickly adapted too, posting concise comments/questions rather than standing up and saying "uh, hi, uh, I'm Chris from Suburb, and uh, yeah, so this proposal, for uh, you know, the walkway thing, oh sorry, I mean, the, um, development application, yeah. So, um, oh, I think, um, it's good. I mean, mostly. But, uh, can we, I mean, can the council, ah, ask the builder, or, you know, the, uh, owner, to, well, I mean, it would be, better, probably, ah, if, ...

Unlike the child imprisoned inside it... yeah, yeah, sure, Tesla's are 100% reliable at detecting even tiny babies inside them and would never allow a repo instruction to override common sense. "don't steal babies".

Although since paws4thot thinks the Europeans can freely modify the software in their Teslas they're presumably not paying BMW for heated seats and will also disable any repo-type instructions. I can't wait for the first FOSS complete Tesla operating system to come out. In Europe, a place I will likely never set foot in. Other side of the planet sounds like an excellent place to test that approach...

The problem in the US is with auto rental companies. Hertz is the one in the news of late. They get their records messed up at times and have had people arrested for stealing a rental car. When the payments are up to date and a credit card is on file.

People are proposing laws.

The concept of "dictionary" something so simple that it was obvious after you explain it to someone. But. But first someone to come up with it and then go through a couple dozen drafts for how entries ought be laid out. It is a critical tool in leveraging the fullest value of language. Most folk rarely need much of a vocabulary, just about 5000 commonly used -- most heavily used -- words in living their typical days.

I'm sure you know that it's possible to explain things in ten hundred words. (There is even a computer to help you in our language.) Ten hundred isn't very many words and we keep thinking of new ones but we know that it can be enough for a lot of things. I'm sure someone has tried figuring out how few words we really need. Five ten hundreds is a lot more than we need most of the time.

I ran these words through the word checker and had to say things in new ways a few times, but I think the readers will all know what I mean.

RE: Neanderthal extinction. There are a couple of new, non-paywalled papers out in Nature about how the now few hundred-odd ancient genomes of modern humans relate to each other and to us.

https://www.nature.com/articles/s41586-023-05726-0

This is about moderns, not Neanderthals, and it's mostly post 37,000 years. There's an interesting bit early on: "Surprisingly, however, none of those pre-40 ka individuals left substantial traces in the genetic makeup of present-day Eurasian populations. The oldest genomes carrying ancestries that derive primarily from the lineage leading to present-day Europeans are Kostenki 14 (from 37 ka, with uncertain archaeological association from western Russia), Goyet Q116-1 (35 ka, Aurignacian-associated from Belgium) and Bacho Kiro 1653 (35 ka, probably Aurignacian-associated from Bulgaria). These data suggest that the genetic ancestries identified in the pre-40 ka individuals analysed so far went largely extinct or were assimilated by subsequent expansions."

One way to interpret this is that the Phlegraean Field eruption in 39,280-ish BC, possibly combined with other global scale disasters in the preceding years, basically wiped out all hominins in Europe. After that, the continent was settled by hominins from the Middle East and Africa, some or most of whom were of hybrid origin.

So it really could have been just bad luck, not sapiens outcompeting Neanderthals on their home turf. Our sapiens and hybrid ancestors may have survived simply because they were outside the death zone.

This isn't what the article is about, and it's worth reading on its own merits.

287 P4 - Are you sure? I know several people who have cars with "collision detection systems", which are very good at braking for blowing crisp packets, parked cars, cars in adjacent lanes... For them to have "detected and braked for a creature in the roadway" would seem to indicate that the detectors had a line of sight to said creature that the nominal controller/operator didn't have.

292 - Er, all I said was that there is no law preventing Europeans from modifying software. I did not say that your specific cites had ever,never mind regularly, been done.

Ah, sorry, I took "yes you can" as meaning it's possible, not merely that the law doesn't forbid it. Viz, the law would need to require that users be allowed to do so, and that law would need to be enforced, for it to be possible.

The "right to repair" kerfuffle in the USA gives many examples of the difference between "not illegal" and "possible".

I was thinking about that when I wrote about self-possessed cars.

But I'm sure that Ford would never allow buggy software to leave the factory, or have buggy software or procedures within the company, and thus the problem of inappropriately (re/de/self/)possessed vehicles will never arise. "my car is possessed" takes on a whole new meaning.

One of the other aspects of self-driving that I haven't seen is the "instruction by a police officer" one. There is no requirement that the instruction be machine-readable at least in Australia, and it is a serious offence to disregard such an instruction. So self-driving cars will need either an exemption from the requirement to obey police, or some way to ascertain the meaning of instructions.

There is plenty of evidence on sites such as briskoda.net that VAG have and regularly do, sell cars with buggy software. There is also evidence of people and companies modifying code and/or hardware to disable or modify some of said bugs.

ilya187
Many, many years ago ... about 1981 - ish ... that happened to me.
I doubt a Tesla would have done any better
I was travelling at about 15mph ( & slowing down ) towards a T-junction, at which I almost always stopped, before turning.
There were parked cars on both sides & I saw a l-h "flicker" followed about half-a-second later, a loud thud. A 9 (?) year old had run out, head-down, between two parked cars.
Very fortunately, he was thrown back, with a greenstick leg fracture, not forward & down, or I would have run over him.
Again, very fortunately, the car some way behind me had a Local Government employee with a transmit/recieve radio in his car - extremely rare in those days. He sent for the Ambulance. About 5 mins later the police showed up, took a look, asked me where he'd emerged - which was confirmed by the other children present. Their next comment was: "You were not going very fast, were you, sir?"
To which I said: "it's narrow, I know this road & I've got an almost-blind T-junction in front, of course I'm going slowly!"
I made a statement later, but that was it.
The child recovered completely, I'm glad to say - he certainly seemed cheerful enough, when I visited him in the local hospital ....

John S
I assume that was brought in by "R" MAGA fuckwits?
What are the chances of it passing? And, even then of it being overthrown in the courts?

H
The Phlagraean Fields is/are still horribly active - it's still a "warm" caldera.
If it blew again, right - Naples would be vapourised, but how bad would the fallout be across Europe?
Though I suppose the answer(s) would vary depending on season & wind-directions at the time.
{ See also: Yellowstone? }

Again, you haven't read "Halting State", have you? That's a major plot point (in a novel I wrote in 2005/06).

There's a brutal joke which I no longer find humorous, about democratic nations losing out to dictatorship.

The flaw in the notion of rule of law, and citizens voting upon who their leaders are, wannabe dictators cheat.

It seems to boil down, down, downwards into "one man, one vote, one time" with the wannabes chipping away at the right to vote.

Not just flaky South American banana republics but now playing out in India, UK, Poland, USA, etc.

Not too late to turn it around, but too many eagerly embracing dictatorship in varying forms.

Those posts about cars that 'self repo', along with tractors which farmers are forbidden to repair themselves and railroads shrugging off safety regulations... all the sorts of trimming away at rights to be expected in dictatorship.

If we slip this particular snare, how "Popular" will any denomination that supported "Christian Nationalism" be? I haven't been a member of any church for over forty years but I still have friends and relatives on the inside, that I wouldn't wish to see harm. On the flip side, if we fail to slip this snare... we will have achieved dystopia.

I strongly suggest you read "Rule 34" before continuing on this path ...

I did, twice I think. Perhaps it's time for the third time though, this pair of books is really enjoyable.

"First, intelligence has been defined from ~1950 to ~2006 as "everything a human can do a computer can't", and AI was considered non-existent. Then with DNNs everything got called AI suddendly. It's not a technical term at this point, it's a marketing term."

The latter two sentences are correct, but the first isn't. Nobody in the field used that definition, though there were (and are) a zillion definitions. The one that I favour and was close to consensus in the period you mention was "Anything we don't know how to program, even in principle." The abilities defined by that have remained remarkably stable since the 1960s, though there have been a few programming breakthroughs.

For example, language parsing, translation and comprehension were not considered AI, but extremely hard linguistic and programming tasks.

Similarly, object recognition and categorisation (as in self-driving cars) were known to be programmable, but doing so well was very hard.

So it really could have been just bad luck, not sapiens outcompeting Neanderthals on their home turf.

My admittedly limited reading on human evolution has given me the impression that the role of luck is usually ignored. Which seems rather strange, given how few specimens we have to extrapolate from.

I'm reading books not research papers, so this may be a biased impression. It's almost a colonialist mentality with some authors: our lineage took over the world, so our ancestors were clearly the most fit, and so they deserved to…

But I'm sure that Ford would never allow buggy software to leave the factory, or have buggy software or procedures within the company, and thus the problem of inappropriately (re/de/self/)possessed vehicles will never arise. "my car is possessed" takes on a whole new meaning.

Two or three "heads of Ford" back they CEO was trying hard to move the company from a parts mindset to a we must figure out how to provide continuous software updates. In other words, if there's an issue with the software in your engine control system you should be able to update it without it being a big deal or replacing the computer itself. But he neglected to keep overall sales up so he was shown the door.

Auto companies (at least in the US) think you should buy a software fix from the "parts department" at a dealer. They mostly have yet to get past the selling of "things" to the selling of systems.

Again, you haven't read "Halting State", have you? That's a major plot point (in a novel I wrote in 2005/06).

I've read it multiple times. Probably I'm unnecessarily cautious, but I figured it was rude to give spoilers on major plot twists on the website of the story's author. So I didn't.

307 Para 3 - In the EU, they do "at least" believe that a software upgrade is/should be part of a "service".

...VAG... have and regularly do, sell cars with buggy software. There is also evidence of people and companies modifying code and/or hardware to disable or modify some of said bugs.

What is VAG?

I had to have a software update to the "entertainment" system on my Honda Civic a year or so after buying it. It crashed in a weird way and got stuck where half the display continuously showed the view of the backup camera and the other half seemed to be legit but the touch screen was totally off.

And about once every few months it gets into a snit and I just force re-boot it. Hold power button for 5 or 10 seconds and you get a prompt.

And if you enter the secret key combination you get to see the Android underpinnings.

The dash / console has never shown such things. I suspect the developers of it are held to a higher standard.

What are the chances of it passing? And, even then of it being overthrown in the courts?

Maybe. But as I've said before, at the start of state legislative sessions all kinds of stupid stuff is introduced as bills. To keep a campaign promise or to be able to use it in a campaign ad in the future. Most of these are just a waste of paper and toner. They get assigned by the adults to a study group and are never heard from again.

But every now and again, one gets real. And craziness ensues.

Someone in Tennessee is proposing a law limiting drag shows. Someone else found a picture of the current governor (I think) who said he'd support such a law in drag at some college party when he was a few years younger. The local news must be interesting just now.

If it blew again, right - Naples would be vapourised, but how bad would the fallout be across Europe? Though I suppose the answer(s) would vary depending on season & wind-directions at the time. { See also: Yellowstone? }

Northern Italy would be a wreak. And a huge chunk of the Mediterranean or Europe would be a mess.

But Yellowstone?

When it next goes it will wipe out an area the size of France, Spain, and Germany combined. And wreak most of the rest of the continental US. Based on the past 3 events that have been studied.

It's big.

Yes. Ignoring that also leads to the mindset "if it didn't happen, it couldn't happen", which is demonstrable nonsense. For some reason, most people seem completely unable to get their minds around probability, and insist on their being identifiable causes for every event.

I saw some purported research that said the primary reason that people became ridiculously rich was luck, followed by starting as merely rich. I believe it. The same is true in academic research; while there ARE geniuses, the vast majority of people who make breakthroughs were lucky, and no more able than some of their colleagues that didn't (but remember that I am talking about a really high standard of ability, here).

Indeed, elementary probability shows that any species that went through a tight bottleneck was lucky to survive. Most such were unlucky.

VAG - Volkswagon Akteingesellshaft. The parent/holding company of Audi, Cupra, Lamboughini, Porsche, Seat, Skoda and Volkswagon cars, Ducati motorcycles, and Navistar, MAN, Scania and Volkswagon trucks. Does anyone not see why people use the TLA?

https://en.wikipedia.org/wiki/Indiana_Pi_Bill

It’s a side-effect of TOS.

Oh, what’s that? TLA Overload Syndrome.

But yes, VAG: one of a not small number of ICE vehicle makers found to have comprehensively lied and cheated in emissions and efficiency testing over a non-trivial time period.

I note that VAG are one of the largest ICE vehicle manufacturers: I speculate (wildly) that they might have been the first to get caught simply because they sell a metric heap of vehicles in the EU where a number of governments seek to apply the EU's strict emissions standards and have different monitoring agencies (so there's no single organization to make an end run around).

Meanwhile, Redwood Materials reports on first year of car battery recycling -- they only processed about 1200 car battery packs (230-odd tonnes) but say they reclaimed 95% of the "lithium, cobalt, nickel, copper, and other metals" in those batteries.

There aren't enough end-of-life EV battery packs for them to ramp up production fast (as EV batteries outlast earlier predictions), but they're already profitable recycling laptop/phone/other LiIon batteries and say the main obstacle is logistics (getting car batteries to the recycling center gate).

So ... this is a long-term trend to watch, but if 95% of lithium in EV batteries can be recycled into new batteries, that makes the lithium power packs in cars a renewable resource rather than a one-time consumable.

PS: to emphasize, this is a report on the first 12 months of a pilot project to see if it was even feasible to scale up phone/laptop battery recycling to cars. One assumes improvements in volume and efficiency are possible.

I doubt a Tesla would have done any better

Probably not, but so what? It is disingenuous to claim "self-driving cars are a danger" on the basis of extreme situations which no human driver could possibly handle.

There is a lot of evidence for that. That case caused most of the major manufacturers to start behaving like ants from a kicked-over anthill. It is possible that there were NO manufacturers that had not lied.

For them to have "detected and braked for a creature in the roadway" would seem to indicate that the detectors had a line of sight to said creature that the nominal controller/operator didn't have.

Since Tesla has side cameras, it is a possibility. I think what is more important is that it has a continuous 360 degree coverage, while a human has only a narrow vision cone at any given moment.

The latter two sentences are correct, but the first isn't. Nobody in the field used that definition, though there were (and are) a zillion definitions. The one that I favour and was close to consensus in the period you mention was "Anything we don't know how to program, even in principle." The abilities defined by that have remained remarkably stable since the 1960s, though there have been a few programming breakthroughs.

My Master degree is in AI, received in 1991, and not once did I hear AI defined as "Anything we don't know how to program, even in principle." The examples you brought up -- language parsing, translation and comprehension, object recognition and categorization, -- were all part of AI as far as my professors were concerned.

In the style of Charlie Stross's book, Why We Shouldn't Build the Torment Nexus...

One can only pray that Ford's lawyer's get this right. (Fat chance, I know.)

Yeah. I think that was sarcasm.

Will there be a quiz at the end of the thread?

I speculate (wildly) that they might have been the first to get caught simply because they sell a metric heap of vehicles in the EU where a number of governments seek to apply the EU's strict emissions standards and have different monitoring agencies (so there's no single organization to make an end run around).

I thought they got caught in the US first. But that was 10 years ago and my memory is fuzzy.

The most surprising thing about the whole affair was that anyone ever expected anything different in the first place. After all, the situation is equivalent to trying to get people to drag different loads on a rope using a single posture, instead of staying upright for the light loads and leaning forwards for the heavy ones. It simply won't happen unless you strap them upright, like the Victorians did with carriage horses.

Rbt Prior & EC
The amazingly stupid mindset you mention ... is the exact opposite of that followed by railway safety professionals { Though, maybe, not in Greece! }
Of course, this stupidity is behind the insanity of "Strict Liability" criminal legislation ... "We've covered every possibility!"
Oh, no you fucking well have NOT, guaranteed ....

I should have probably clarified that I was talking about the first half of the period 1950-2006, before the term 'AI' was used solely as a marketing label. See Olivier Galibert (#279), who is dead right on that aspect, though not on the date of the change, which was c. 1980.

I was close to people working in that era at the time the change took place, and the reasons that so much existing work was labelled as AI was (a) academic bullshitting to increase the prestige of unexciting work and (b) bullshitting to encourage funders to give large amounts of money for such work.

The eminent and respectable academics who had been doing such work for decades were throroughly disgusted, I can assure you. Edinburgh University started working on AI in 1963, but were fully aware that it was a goal and what they could do was not yet it. Both of those statements were made to me in person by eminent people in the field.

A human has over 120 degrees of lateral field, and our peripheral vision has evolved to pick up motion. That's ample for a moving car, though I agree not enough to avoid being run into from the side. In the UK, loss of peripheral vision can cause you to be refused a licence.

All I can say is my own experience -- several times over last 2 years my Tesla had reacted to a pedestrian before I was aware the pedestrian was there. And I am an extremely safe driver, according to my insurance company.

Since we've passed 300, allow me to muse on recent news about US(*) agencies (eight of them) opining about the origins of COVID. The new news is that the Department of Energy, apparently meaning Lawrence Livermore's Z Division(**), has changed from "we dunno" to low-confidence "looks like maybe a lab spill".

In the wake of that, FBI Director Wray was on Fox News and said,

“You’re talking about a potential leak from a Chinese government-controlled lab that killed millions of Americans and that’s precisely what that capability was designed for.”

As one normally parses English text, the antecedent standing in apposition to "that capability" would seem to be a megadeath bioweapon. But a few sentences back Wray was talking about FBI forensic activities relating to biohazards. So maybe that's what he meant and just had a slip of the mouth. Would be nice to get clarification.

(*) Is anybody else worrying about where it came from. Should anyone care?

(**) Los Alamos has a similar group -- I wonder if they have also opined. LLNL and LANL don't always see eye to eye.

Well, the American right is currently telling their followers that apparently the FBI has proved that the release of Covid was an attack on America using a weapon of mass destruction… and that this is Biden's fault…

I know. Clearly the best way to launch a bioweapon against America is by releasing it at a Chinese market, Trump decided it wasn't a problem, etc etc. But how many of the fanatics who think Fox is becoming too woke and left-wing will even attempt to understand that, when their elected leaders refuse to?

Obviously I don't know the specifics of the current Greek rail tragedy, but I must point out that the entire rest of Europe absolutely demanded that Greece eviscerate every public program to appease the debt gods (said debt having been incurred through buying things like military hardware from the rest of Europe).

The wildly obvious fact that austerity NEVER WORKS is a part of a larger scale stupidity. The other plain fact is that when you force deep cuts to things like railway safety, you have decided that a certain number of deaths are quite acceptable to protect your bondholders (banks).

As has been mentioned on here a few times, the purpose of a system can be found in its results. In the case of the current financial system, the purpose is to kill as many little people as necessary to make damn sure the banks get to keep their money.

In the instance of the Greek train crash, externally imposed austerity can be directly linked to those many deaths.

In the wake of that, FBI Director Wray was on Fox News and said

My take is that this is 101% pure unadulterated bullshit.

a Chinese government-controlled lab ... killed millions of Americans

Yeah well, it also killed millions of Chinese.

The political polarization around COVID19 and public health in recent years in the USA is terrifying, as is the tendency of the executive to meddle with appointments to intel services to deliver the reports they want to hear -- Bush II was doing it as well, this isn't new. More to the point, one would think that if the Chinese government had a bioweapon-related lab leak they might have been slightly better at preventing their own civilian casualties (or maybe deploying a vaccine or three).

I think this is just ammunition for a future culture-wars-alternate bullshit campaign, flooding the zone with crap (as Steve Bannon puts it).

and that this is Biden's fault…

Yes, blame Biden for an event that happened before Joe was elected President, that totally makes sense, like the loony-tunes who blame Obama for 9/11.

I mean, there is bullshit, and then there is this species of rancid excrement and I think we here are all old enough to remember when these events actually happened?

TLDR is, anyone spreading this crap has a political agenda and an axe to grind and is lying through their teeth.

321 - So we go from "$vehicle makes random braking moves" to "Tesla brakes so that creature runs into car"... Hardly an improvement I suggest.

332 - There is more than one way of interpreting that.
1) The pedestrian was actually in danger of colliding with a moving vehicle.
2) The pedestrian intended to stop near your vehicle, and it over-reacted.
3) You don't actually scan as wide a cone of vision as you think you do.

335 Para 1 - 13 of 27 types in the Hellenic Air Force's current OrBat were sourced in the Americas.

Charlie @ 235: Marvin Minsky set neural networks back decades when in the late 1960s he came up with a clear proof (not disproven since, either!) that a perceptron-based network was computationally too weak to learn an XOR system and therefore wasn't capable of general-purpose computing.

Yes, I remember that quite clearly, as I was involved in an amateur attempt to build some neural networks back in the 80s. (We didn't get anywhere because we were trying to do something on 8 bit CPUs in BASIC, with very limited knowledge of linear algebra).

I remember pointing out that the only way a human could do a parity check was to count up the black squares and decide whether the total was odd or even, so the fact that perceptrons had the same limitation seemed to suggest that the human visual cortex was working on similar lines.

This image includes a pair of nested spirals, the point being that a perceptron can't figure out whether the shapes are open or closed (its a variation of the parity problem). But short of carefully tracing the shapes with your finger, you can't see it either.

David L @ 311: Someone in Tennessee is proposing a law limiting drag shows.

Thinks... pantomime!

For those outside the UK, this is a really weird thing we have here. A standard part of Christmas is taking the children to the "panto". Here is a trailer from one. And yes, that is a middle-aged man in drag ("the dame"). Also the male romantic lead is often played by a woman ("the principal boy"), opposite another woman playing the female romantic lead. They generally kiss at the end. Its completely formulaic, and lots of fun. For most children its the only theatre they know until they get thrown into the deep end of Shakespeare in secondary school. Productions run the gamut from big-budget West End featuring famous names down to a local amateur dramatics group putting on something at the village hall. I've seen both, and I actually prefer the am-dram version.

(Its descended from the Italian Commedia dell'arte

Those Tennesseeans would have their minds blown: a drag act alongside two women making out specifically aimed at children in the 5-12 age range! And this is the British idea of good, wholesome family entertainment.

Are you saying the The Laundry operatives couldn't possibly be involved in something so bizarre? Retrocausality is a step too far?

pwas4thod @ 296: For them to have "detected and braked for a creature in the roadway" would seem to indicate that the detectors had a line of sight to said creature that the nominal controller/operator didn't have.

One of the scary things about modern cars is the thickness of the front windscreen pillars. There is good reason for this: those pillars are what stop you being crushed to death in a rollover. But they also block out a sector of your visual area. And if something is on a collision course it means its on a constant bearing. So the child running into the road who is hidden by the pillar stays hidden until its too late.

Having a computer looking through a camera which doesn't have that blind spot sounds like a really good idea to me.

I am afraid that this is the USA getting its excuses in first, as it ramps up its hostility to China. That does not fill me with joy.

Yes, but the vast majority of the cases of drivers not seeing pedestrians and cyclists are simply that they are not looking for them. I try to look for them in preference to cars, and usually succeed - this does mean that I very occasionally have to do an emergency stop to avoid running into the car in front, but I regard that as better than risking a vulnerable road user. I have lost track of the number of times a driver has said "That pedestrian/cyclist was invisible / jumped out at me" and I (as a passenger) had seen them over 10 seconds earlier.

The pillars probably are the second most important reason.

TLDR is, anyone spreading this crap has a political agenda and an axe to grind and is lying through their teeth.

Well, yeah. "Coming from the American right" does kinda imply a political agenda. Not to mention lying.

One of the scary things about modern cars is the thickness of the front windscreen pillars.

Trying to remember which car manufacturer I saw in a newspaper article was working on building curved high-def screens into the A-pillars that would project a camera view of whatever was blocked by the pillar from the driver's eyeball point. Thereby rendering them sufficiently transparent for collision avoidance.

(It's also a known problem for fighter jets -- the F-16 gets around it by having the support arch for its windshield behind the pilot's head, but in a head-on approach an off-bore adversary who can sneak up on your blind spot at a closing speed of a thousand knots is a serious threat.)

Gleiwitz, 2023
Putin's "excuse" for a dangerous escalation, or a new attack?

RocketJPS @ 335
You will not find me disagreeing with anything at all you have said, there ....
It's "just" more of the sme stupidity.
- see also EC @ 343.

Paul
Along those lines ....
"Where's the Evil Villain(ess)?"
He/she's BEHIND YOU!
Etc Ad nauseam - oh dear.

"I am afraid that this is the USA getting its excuses in first, as it ramps up its hostility to China. That does not fill me with joy."

This is the Murdoch Cinematic Universe cranking up the salience up anti-Jyna conspiracy theories in the MAGA fever swamps. Possibly this is purely selfish on Rupe's part, just looking for something to distract the monster from the Dominion lawsuit or generally keep it hopped up on outrage to juice the ratings; possibly it's actually a focused operation, part of Murdoch's general coordination with/puppet-mastering of the GOP to help set the news agenda in support of some other objective.

Whatever the deeper story behind why this is being pushed, the item broke in the WSJ over the weekend and then was given a hard hype cycle push on Fox News which laundered it sufficiently to get it into the more mainstream outlets, so this is a fairly typical Murdoch media shaping exercise.

Of the eight divisions of the stateside IC who have looked at the question of COVID's origin - two have now come down on 'lab leak', four have come down on 'natural breakout' and two haven't been able to make a determination; all of these assessments are tagged as 'most likely' and 'low confidence' (ie IC-speak meaning 'this is what best fits the facts we have, but it is based on extremely hazy and fragmentary intel') and the DoE aren't saying what caused them to flip their assessment from 'natural breakout' to 'lab accident'; so this would be pretty much a non-story without the energy brought to it by having a festering gaggle of flying monkeys who have been artfully curated to believe several impossible things each prime time. As such, if this is being promoted as part of a deeper plot, this looks to me like it is aimed at a domestic target rather than a foreign adversary.

Of course that doesn't stop it from potentially doing splash damage to the foreign adversary (or US govt policy towards them), that's part of why Murdoch playing these sorts of games is so dangerously reckless.

this looks to me like it is aimed at a domestic target rather than a foreign adversary.

It reminds me of "Iraq definitely has weapons of mass destruction, we just can't show you the evidence".

One of the scary things about modern cars is the thickness of the front windscreen pillars.

Hmmmm. Interesting comment.

I have a 2008 Tundra Truck (big one that it is) and a 2016 Honda Civic (smallish one that it is) and I have no issues seeing around the pillars. I might if I only had one eye but as it is I see around them just fine.

Much better that on my cars from 20 to 60 years ago. Especially in the days of "vent" windows.

"That pedestrian/cyclist was invisible / jumped out at me"

I have a big beef with people walking on the side of a street (without what we call sidewalks) wearing totally inappropriate clothing. All dark with nothing light at dusk or dark.

Plus, while it irritates some people if there is no on coming traffic I'll drive more down the center of the road to avoid being close to cars on my side to avoid doors opening and people/children from popping out from the other side of a car. In the burbs.

Yeah, you can suggest pedestrians add reflectors or lights, but fundamentally as the guest on the road* you're the one with the obligation to avoid pedestrians. It's really annoying to have a vehicle capable of 100mph but be limited to 50mph, or even 20mph, purely by your desire not to kill someone, but that's how it is.

OTOH a surprising proportion of walking and running shoes these days have reflective bits on them. The days when I had to sew or glue on strips of retroreflective "cloth" tape are long gone.

* even the US doesn't have a "right to drive" the same way it has a "right to travel" or a "right to walk".

Oh, good, we're over 300. Meme sighting: now I understand why you like a Mac, Charlie. "It's bad luck to mention a Macbook inside an office. You have to call it 'The Scottish Laptop'"

YMMV, but for me, having been the pedestrian or cyclist as an adult makes it easier to notice pedestrians & cyclists. A possible bright spot for us caught up in the reactionaries quest for their "Big Rock Candy Mountain" is more people will experience the road first as cyclists & pedestrians and carry a little empathy over if they later become motorists.

Strunk and White are wrong about almost everything.

Ha! That article is gold.

S&W has an unlikely following in the (mostly fiction) writing community in Oz in recent years. Stephen King's book on writing, which endorses S&W strongly, is popular here and seems to provoke a fresh surge of interest with each re-release. I suppose it's an adjunct to the need to target the US reading market if you want to make a living writing in English, but it comes with a touch of "if it's good enough for Stephen King..." about it.

I have no issues seeing around the pillars. I might if I only had one eye but as it is I see around them just fine.

I find with the little Mazda, the pillars are not the problem so much as the rear-view mirror. I'm 6'1-and-a-bit, with the driver's seat adjusted as low as it will go and the backrest tilted as far back as I'm comfortable to tilt it, the mirror still occludes a significant part of my field of view. It's exactly the location where, approaching T-intersection from the stem, a vehicle coming from the left and turning right would be be converging, assuming they are slowing for their turn at a similar rate to me (Americans might need to invert left and right in that sentence for it to make better sense). I've had a couple of near-misses, including one where the other vehicle stopped to let me through, even though I was clearly in the wrong. I have therefore learned to nod or otherwise move my head to look around it, which must look peculiar to others, especially now it's a habit and I do it in other vehicles where this problem is not necessarily applicable. I suppose it's not onerous, it's less movement than a proper shoulder check, and I guess it probably helps with other blind spots like pillars. Anyhow, it's one of the downsides to a small car. Bicycles don't have this problem, at least not usually.

Maybe Biden is trying to send China a signal... You might not want to mess with us on the whole Russia thing, because we might just mess back. Something along the lines of "Don't push us or we might decide those lies about you are true."

If so, it's very short-sighted in a whole lot of ways, but so are most American policies.

Yeah, I find myself ducking and weaving while driving for much the same reason. Unless I'm driving a truck where the viewing angles are usually better even though the remaining blind spots are bigger (things next to the doors... invisible unless very tall. Things behind the vehicle just not visible at all)

But most of my travel time is on a bicycle, then walking, then on trains (ordered by distance I go further by train than on foot, though). Except for the interlude where I was dating a habitual motorist my time spent in cars/vans etc is normally zero per year. Which in circular fashion means I'm a terrible driver so I tend to avoid doing it.

340 - Personally, my favourite pantomimes tended to be regional level, and might involve people like cast, directors, producers and/or writers from the 7:84 Theatre Company, Jack Milroy, Jimmie Logan...

342 - True, as far as it goes. I can't remember a link to the analysis, but I've seen a case where a stationary at time of impact motorcyclist was responsible for a collision with a signalling and turning truck because the motorcyclist rode straight down a blind arc to the point of impact.

This seems a relevant danger: "Oceania Has Always Been at War with Eastasia: Dangers of Generative AI and Knowledge Pollution"

https://www.bigmessowires.com/2023/03/01/oceania-has-always-been-at-war-with-eastasia-dangers-of-generative-ai-and-knowledge-pollution/

What’s fascinating to me is that as recently as a year ago, virtually nobody was talking about this risk. It wasn’t on anybody’s radar. We worried about AI putting people out of work, or maybe about Skynet. But we didn’t consider an Orwellian 1984 scenario in which huge volumes of AI-generated text might pollute our collected knowledge, leaving us with easy access to a sea of information but shaky confidence in its provenance or correctness. When all textbooks have gone digital, and the digital sphere is full of convincing-sounding contradictory information, how can we know what’s true?

"I speculate (wildly) that they might have been the first to get caught simply because they sell a metric heap of vehicles in the EU where a number of governments seek to apply the EU's strict emissions standards and have different monitoring agencies (so there's no single organization to make an end run around).

I thought they got caught in the US first. But that was 10 years ago and my memory is fuzzy."

Wikipedia agrees with you

CharlesH: I think that's leaving out part of the story. The crucial issue for Grevan and the kids colonizing the new planet wasn't whether they could eat the food or not. It was whether they could break free of the Dominator that was sent along to mind control them if they tried to stay there. That is, it was a conflict between them and Central Control. And Priderell's ability to eat natural foods and not die enabled her to join them. Dietary unspecialization seems to be a symbol for general independence.

Wikipedia agrees with you

Doesn't matter, though: the basic postulate (tight emissions regs, non-compliance exposed through testing in multiple jurisdictions with competing enforcement) stands.

It is, however, substantiating evidence that when the EU and USA agree on something (vehicle smog emissions, for example) it tends to make everyone else fall into line (those being the two most lucrative markets on the planet, until India and China finally reach parity).

My main obstacle for cycling now is wanting to take dogs along. I know I've talked here before about wanting to build a DIY cargo bike, and at other times about wanting to convert my old MTB to electric with a mid drive and internally geared hub. These days I'm much more serious about just saving up and buying one of these and having done with it. Even if the "for the Australian market" adaptations are a little bit arse (specifically, replacing the regenerative roller brakes with hydraulic disc brakes... I'm not sure which state's laws that is being done to appease). The "gel saddle" in place of the nice Brooks B67 in the pictures just sounds like adding insult to injury there, but that one is easier to swap out than the brakes.

I suppose I still have questions that are right up the alley of several commenters here, but I think we've oversaturated cycling as a topic and it's just better all around not to keep going there.

There's now a bit more on official findings concerning Havana Syndrome:

https://www.cia.gov/stories/story/statement-cia-director-burns-on-ic-assessment-ahis/

https://www.dni.gov/index.php/newsroom/press-releases/press-releases-2023/item/2361-dni-statement-on-the-intelligence-community-assessment-on-ahis

The latter has this [IC = Intelligence Community; ICA = Intelligence Community Assessment; AHI = Anomalous Health Incident, aka Havana Syndrome case]:

Based on the latest IC-wide effort, which has resulted in an ICA that will be issued today, I can share with you that most IC agencies have now concluded that it is “very unlikely” a foreign adversary is responsible for the reported AHIs. IC agencies have varying confidence levels because we still have gaps given the challenges collecting on foreign adversaries — as we do on many issues involving them.

As part of this review, the IC identified critical assumptions surrounding the initial AHIs reported in Cuba from 2016 to 2018, which framed the IC’s understanding of this phenomenon, but were not borne out by subsequent medical and technical analysis. In light of this and the evidence that points away from a foreign adversary, causal mechanism, or unique syndrome linked to AHIs, IC agencies assess that symptoms reported by US personnel were probably the result of factors that did not involve a foreign adversary, such as preexisting conditions, conventional illnesses, and environmental factors.

Since we're well over 300, I have a question for everyone: What sciences currently have a method of proving that Earth is more than 6000 years old? I know that anything which deals with genetics can prove this, (biology, botany, medicine, zoology, ornithology, etc.,) as can astronomy, paleontology and geology, but what else can offer satisfactory proof of Earth's age as well-over 6000 years?

My "maybes" include geography (as relates to geology,) archeology, (tree rings go back about 12,000 years,) physics (as relates to astronomy at least) and possibly linguistics (they might have a clock about how language changes) but what else can manage the proof? Can anyone else confirm one of my maybes or add something new?

Also,

https://www.dni.gov/files/ODNI/documents/assessments/Updated_Assessment_of_Anomalous_Health_Incidents.pdf

As awful as the recent mega-train wrecks have been -- India, Austria, Greece, US, etc -- it has made it increasingly obvious the need for regulatory enforcement and an end to toothless tiger inspections. Because as bad as these wrecks have been, the potential for something much (much!) worse is literally around the bend. But apparently all the wrong lessons are being learned such as better bullshit spinning by spin doctors and smoother soothing words by amoral CEOs. Ugh.

I so wanted to believe the US has begun clawing itself way out of the stupid pit... and now this?

"20,000 people may have been exposed to measles at a large religious event in Kentucky"

given how few people understand the potential damage possible from measles it will be a shock when there's crippling injuries to adults and birth defects inflicted upon newborns... all due to wack jobs deluded into fearing vaccines more than viruses... and has anyone got a reliable sourcing for whether there is indeed another variant of covid arising? I've noticed there's next to zero about that and not much attention about increasing numbers of folks being ground down by 'long covid'

theguardian.com/us-news/2023/mar/03/measles-outbreak-kentucky-religious-revival-event-exposure

Remember that you can't actually prove scientific hypotheses, only disprove -- or really, call into question.

How can we know what's true?

In the context of "place your bets", perhaps pre-2022 non-fiction texts are the place to put your money. Or buy up used book stores. Before the flood of plausible but completely wrong books hits the market.

You can't.

There's no convincing way of disproving the Omphalos Hypothesis -- or its less overtly religious, modern, variants, the Simulation hypothesis and the you-who-are-asking-the-question are a Boltzmann brain hypothesis.

However, the degenerate case for all of these ideas is that they are isomorphic with solipsism and deserve exactly as much respect (i.e. none whatsoever).

Since we are past 300...

Most of you probably know who Ted Chiang is. Chiang may or may not be aware of Charlie's "corporations are alien invaders" notion, but he seems to have come up with something very similar back in 2017 (or earlier):

https://www.buzzfeednews.com/article/tedchiang/the-real-danger-to-civilization-isnt-ai-its-runaway

Sigh. Of course I understand that. But some hypothesis are much better than others, and eventually a really good hypothesis graduates to "theory," which means it comes as close to being scientifically proven as possible.

Sighs again

I think it was very obvious what I was asking and what I wasn't asking. And yes, I understand all these possible ontological and philosophical issues, but that's not the question I'm asking.

What I'm more concerned with is this: Fifty years ago, Christians trying to prove the "6000 years" hypothesis were mainly arguing against the biological sciences and the theory of evolution. But at this point it seems like every science has gotten to the point where it has at least one "proven" theory, often multiple proven theories, which points to a much longer universe than 6000 years, and I'd like to catalogue them. That's what I'm interested in.

Re: 'No evidence for any hominid domesticating or even taming any species while the Neanderthals were extant, so far as I know.'

Based on the slew of aww-type Twitter vids I get everyday: Put any two youngsters (regardless of species) together and they'll bond. Often, for life.

I think we need to consider the full range of demos of humans/critters, circumstances, intentions/motivations and actual relationships.

There's no convincing way of disproving the Omphalos Hypothesis -- or its less overtly religious, modern, variants, the Simulation hypothesis and the you-who-are-asking-the-question are a Boltzmann brain hypothesis.

Great quote from the link: "God essentially created two conflicting accounts of Creation: one in nature, and one in the Torah. How can it be determined which is the real story, and which is the fake designed to mislead us? One could equally propose that it is nature that presents the real story, and that the Torah was devised by God to test us with a fake history! ...

"One has to be able to rely on God's truthfulness if religion is to function. Or, to put it another way—if God went to enormous lengths to convince us that the world is billions of years old, who are we to disagree?"

The problem with the Omphalos hypothesis in Christian terms is that it asks us to believe that God commands us to freely believe he exists and to believe the statements and follow the commands in his book even when they contradict eaceh other, threatens to punish us with eternal damnation if we do not, and also created a world that shows no evidence that he did any of this, and in fact shows evidence that parts of the key historical record disagree with a text.

This asks us to believe that God lies, routinely and elaborately, in ways that make it easy for us to condemn our souls to hell. Logically, therefore, the god of the Omphalos Hypothesis is Satan. Since I know of no commandment in the Bible to believe it is literally true, combined with skepticism about Genesis going back at least to St. Augustine, I think the Omphalos Hypothesis is more safely avoided.

What sciences currently have a method of proving that Earth is more than 6000 years old? I know that anything which deals with genetics can prove this, (biology, botany, medicine, zoology, ornithology, etc.,) as can astronomy, paleontology and geology, but what else can offer satisfactory proof of Earth's age as well-over 6000 years?

Prepping for or in a debate with one or more YEC adherents?

Based on my experiences 10 years ago, the areas of science you mentioned will be hand waved away as theories.

But geology, ice cores, volcanic magnetic residue, impact craters, etc... are hard facts that the YECs want to hand wave away but they can't. And many times they will have a hypothesis that supposedly explain one or more sets of "facts" but these don't work together. In most cases they contradict each other. Or fail when scratched below the surface.

Much of the YEC comes from a religious basis. And there is a web site named about answering in the first book of the Bible. (I'll not put a link here to keep Charlie happy and this web site flooded with crazies.) Basically they say, we know the answer and any proposal that doesn't lead to this answer must be false and therefor wrong. It used to be hard to find on their site but they have a page or few of things they used to claim but now say "never mind", we were wrong, don't use this argument.

In the end, at least in the US, this has become one of those YEC is correct, anti-YEC is evil thus anyone not believing in YEC is evil and can be ignored.

I have a question for everyone: What sciences currently have a method of proving that Earth is more than 6000 years old?

Short answer, pretty much all of them. Check out https://rationalwiki.org/wiki/Evidence_against_a_recent_creation for a long, and somewhat snarky, list of these. It may be of interest just for being sorted by time.

I’ll confirm astronomy (many ways) and geology (many, many ways) but don’t know enough about linguistics to say with any confidence.

An odd quirk that I might not have thought of: when created, diamonds are made of carbon. (Duh.) But if you leave a diamond exposed to the environment for thousands or millions of years, nitrogen atoms can slip into the carbon matrix. A diamond lattice doesn't really like nitrogen, so to first approximation diamonds soak up no nitrogen atoms - yet the most common impurity in natural diamonds is nitrogen. This is used as an easy way to differentiate between natural and manufactured diamonds.

But at this point it seems like every science has gotten to the point where it has at least one "proven" theory, often multiple proven theories, which points to a much longer universe than 6000 years, and I'd like to catalogue them. That's what I'm interested in.

There are multiple books and web sites which have done such. The peak of this activity was 10-15 years ago. And a lot of it has religion all mixed in. Yes, there are a lot of religious folks who think YEC is nuts. But they are not in the pulpits. Check out Francis Collins.

If it's OK with Charlie I can dig up a few links and post them. But these links might attack a crowd that he doesn't want to deal with. They can be bad enough going after religious blogs that do not ascribe to YEC.