Soshul meedja.

[Moose-tache]

chris_notts: "True, but isn't this just a further argument against the Chomskian school?"

We don't need more arguments against Chomsky. His school of linguistics is absurd on its face and always has been.

You say this, but it hasn't stopped him and his ideas dominating much of the field for decades, despite the fact that everything we know about how evolution and cognition work would suggest he was talking rubbish. It's depressing how insular and anti-empirical much of linguistics has managed to be.

Chomsky's ideas on UG are about 1% of his work on linguistics and 0% of the useful part. Don't talk rubbish yourself.

Who said anything about UG? Oh, going back with crtl+F it does seem that chris_notts mentioned it. I certainly wasn't talking about UG when I pointed out that there are people writhing on the floor in Pentecostal church revivals who have contributed more insight into linguistics than Noam Chomsky, and I didn't realize that was what you two were arguing about.

[Rounin Ryuuji]

On that note, if I remember right, Glossolalia doesn't actually use the linguistic processing parts of the brain. It's a very interesting, and also vaguely disturbing, phenomenon.

[rotting bones]

1 yes, it is much harder and 2 yes, we don't know how to do it, but the analogy still stands: it seems very likely that instead of modeling every axon of every neuron, you instead abstract the system into, I don't know, parts, and model what those do: the emulator has like, I dunno, a registry: it's not modeling the individual electrons as they would travel through the wires of a snes. admittedly, this is first principles, it could be that minds are too complicated to emulate at higher abstractions than the atom, but I see no reason to think so.
3 I don't think the brain uses quantum effects anymore than the climate does, and anyway, those can be modeled too!

There are scientists who think quantum processes are important for consciousness. I agree they're probably wrong, but that still doesn't settle how fine-grained your simulation needs to be. Protein folding is still a hard problem.

Also, the notion of an uploaded cloud workforce is... AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

We just need AI to reach the point where it realizes that it does not need humans...

You don't need sentience or sapience to get work done. You can potentially automate your entire workforce with non-sentient AIs. Plus, non-sentient AIs won't ask for a salary.

can't speak for you, but the way I know you have feelings is cause i'm very similar to you (made out of meat, same amount of appendages plus minus one or two, same organs, etcetera) and since I know I have emotions that affect mybehavior in this and that way, I guess that you do too. I don't know this apodictically, but I don't know anything apodictically, all is guesses.

But did you apodictically deduce that you didn't apodictically deduce anything else?

[rotting bones]

This isn’t the argument. The argument is that a superintelligent AI just has to be fixated on any goal. If ‘preserve humanity‘ isn’t its priority, then there’s a good chance it won’t preserve humanity.

(The obvious answer to that is that then we should figure out how to make ‘preserve humanity’ a priority for it; that’s basically what the AI alignment people go on about.)

Why would we ever put AIs with independent goals in charge? I think non-CS people take the marketing buzzwords "Artificial Intelligence" way too seriously.

The idea here is that AIs may be able to create and control inconceivably advanced technology, which in practice boils down to ‘nanobots’. (This is also in Alexander’s post.) Personally, I disagree that AIs will be able to create nanobots that readily, if at all.

I have a more fundamental objection. You obviously can't make any headway in the world without intelligence, but I personally think that money makes a much bigger difference under our current economic system. You don't need a super-intelligent AI to wreak havoc if it can find an angel investor to give it tons of cash. In fact, the intelligence need not be artificial. Humans do that already. Conversely, a super-intelligent AI won't be able to do anything unless it gets access to someone's bank account.

[rotting bones]

Multiple pages into to omnidirectional epistemological slap fight, and my original conclusion remains undisloged: If you think the logical functions performed by a computer cannot be a form of consciousness, you do not know what consciousness is made out of.

Dan Dennett put it best many years ago, when he summarized his critics' argument: The mind has to be unknowable. Therefore, any process a scientist can explain, by definition, cannot be the way the mind works.

Honestly, I'm not sure humans are "intelligent" since we're also finding the combinations that our environment shaped us for. I'm a skeptic regarding the traditional concept of "intelligence". Nevertheless, I don't believe this means we can make sentient AI just by writing code. That would mean you can create subjective feelings by entering a sequence of operations into a calculator. Although that is possible, I think it's much more likely that the brain contains organs that generate feelings, and it's impossible to have real feelings without organs that generate them.

My guess is that "intelligence" (in the AI sense) is orthogonal to "sentience". And intelligence in the traditional sense doesn't exist.

[rotting bones]

But that work had a clear advantage present work did not: it could be understood by a human. It now seems likely that we could produce general AI without, in fact, learning anything about intelligence.

The incomprehensibility of neural networks has been greatly exaggerated. Here's an old book on improving learning by meddling with the internals of the network: https://github.com/Honei/bookshelf/blob ... .Trade.pdf I've read papers where they literally go into an image classification network and trace every single path that leads to a positive classification just to make a point.

There are fads in AI as there are in linguistics, so there may be a tendency to dismiss non-neural-net-based work.

We don't use deep learning for no reason. Neural networks work very well on large and complex datasets. There has also been work on rendering them Interpretable: https://arxiv.org/abs/2103.10689 (The Interpretable Machine Learning book I posted earlier uses non-deep learning models.)

[Moose-tache]

Proposed Axiom of Impossibility of Artificial Intelligence

1) As long as there are Filipino teenagers willing to work for $5 a day, that will always be easier than AI. You get turing-complete intelligence, and someone else eats the manufacturing and R&D costs.

2) When there are no more Filipino teenagers willing to work for $5 a day, there is no longer any market for AI. Our global economy cannot function without income disparity, not to mention all the supply chains that would stop dead without cheap labor.

In other words, we can only want Turing-complete robots if we have leisure time to sit around and make sexy chat with anime girls while the global south peels our grapes for us. But the existence of the global south allows us to outsource all the troublesome work that AIs would do.

This is not the same as the "industrialization trap" people talk about with, say, early modern China. That was an avoidable dilemma, as seen in Europe, which managed to increase both automated productivity and wages through the clever strategy of going to other people's countries and shooting at them. That won't work here, because the fact that Filipino teenagers are poor is not a bug, but a feature. We depend on them having less money than us, in order to make the flow of capital and goods function.

[Travis B.]

Proposed Axiom of Impossibility of Artificial Intelligence

1) As long as there are Filipino teenagers willing to work for $5 a day, that will always be easier than AI. You get turing-complete intelligence, and someone else eats the manufacturing and R&D costs.

2) When there are no more Filipino teenagers willing to work for $5 a day, there is no longer any market for AI. Our global economy cannot function without income disparity, not to mention all the supply chains that would stop dead without cheap labor.

In other words, we can only want Turing-complete robots if we have leisure time to sit around and make sexy chat with anime girls while the global south peels our grapes for us. But the existence of the global south allows us to outsource all the troublesome work that AIs would do.

This is not the same as the "industrialization trap" people talk about with, say, early modern China. That was an avoidable dilemma, as seen in Europe, which managed to increase both automated productivity and wages through the clever strategy of going to other people's countries and shooting at them. That won't work here, because the fact that Filipino teenagers are poor is not a bug, but a feature. We depend on them having less money than us, in order to make the flow of capital and goods function.

The thing is capitalists also need (comparatively) wealthy Westerners (or like) to spend more money on products than the capitalists originally paid Filipino teenagers for capitalists to make a profit, and without a profit the capitalists cannot exist. If the capitalists transfer all the jobs to the Third World to be done by poor Third Worlders, there won't be any wealthy Westerners to buy said products, so the capitalists will not make a profit, and thus undo their own economic system.

Same thing goes with AI - if the capitalists replace all the workers with AI's that won't be paid at all, there will be no workers to buy products in the first place.

[foxcatdog]

ai consumers when

[Moose-tache]

The thing is capitalists also need (comparatively) wealthy Westerners (or like) to spend more money on products than the capitalists originally paid Filipino teenagers for capitalists to make a profit, and without a profit the capitalists cannot exist. If the capitalists transfer all the jobs to the Third World to be done by poor Third Worlders, there won't be any wealthy Westerners to buy said products, so the capitalists will not make a profit, and thus undo their own economic system... Same thing goes with AI - if the capitalists replace all the workers with AI's that won't be paid at all, there will be no workers to buy products in the first place.

This is partly right, but the need isn't for wealthy buyers. The need is simply for an imbalance. Aviva Chomsky, the only good Chomsky, wrote extensively about this and the migration of the textile industry. But there is a slight difference when it comes to AI. Owners only need buyers if they need cash income to run production. Once the factory is completely self-sufficient, you don't need buyers. You just lay down with your mouth open under the part of the factory where wealth comes out. Turing-complete intelligence was the last thing that couldn't be automated. Once that goes, there is no way to entitle yourself to a wage, because there is literally nothing the owner needs from you. There is also no way to entitle yourself to any of the stuff, because you don't have anything the owner wants. Like I said, this isn't going to happen because of the need for Filipino teenagers. Not that the Filipino teenagers are actually necessary for productivity. They are necessary for structural and ideological reasons. Once they disappear, the economy stops working, and their existence fuels the innovation of technologies like AI. Most of the advances in software over the last ten years have been based on the distribution of Filipino teenagers (Fiver), or local equivalent (Uber). The imbalance and the technology form a sybiosis.

[Raphael]

Proposed Axiom of Impossibility of Artificial Intelligence

1) As long as there are Filipino teenagers willing to work for $5 a day, that will always be easier than AI. You get turing-complete intelligence, and someone else eats the manufacturing and R&D costs.

2) When there are no more Filipino teenagers willing to work for $5 a day, there is no longer any market for AI. Our global economy cannot function without income disparity, not to mention all the supply chains that would stop dead without cheap labor.

In other words, we can only want Turing-complete robots if we have leisure time to sit around and make sexy chat with anime girls while the global south peels our grapes for us. But the existence of the global south allows us to outsource all the troublesome work that AIs would do.

One possible flaw in your argument is that it implicitly relies on all capitalists being good at planning ahead. What if a capitalist corporation that didn't have the foresight of hiring you as a business consultant develops a market-ready AI because they couldn't be bothered to think through the things you just explained?

A second flaw: Your first point relies on people always going for the cheapest option, in all contexts. Of course, if that would be true, all the world's luxury brands would have gone bankrupt long ago. It's not that difficult to imagine a future in which having expensive AI robots in place of cheap human domestic workers becomes the latest in status symbols for the rich.

This is probably more likely as a result of the fact that a lot of human beings have, for a very long time, been fascinated by the idea of AI. It's likely that there's at least some overlap between "people who are fascinated by the idea of AI" and "people who have ridiculous amounts of money". And people in that overlap might well be interested in buying expensive AI robot domestic workers to replace their cheaper human domestic workers.

[zompist]

This is probably more likely as a result of the fact that a lot of human beings have, for a very long time, been fascinated by the idea of AI. It's likely that there's at least some overlap between "people who are fascinated by the idea of AI" and "people who have ridiculous amounts of money". And people in that overlap might well be interested in buying expensive AI robot domestic workers to replace their cheaper human domestic workers.

Sure, but domestic work is a pretty small sector. 2.2 million in the US, but that's at all levels of society-- I can't find figures on the 1%.

Rich people like novelty, but they like safety better. Would they entrust their children to a robot? Maybe their gardens, but I doubt that techbros have deep fantasies about having robot gardeners.

[Raphael]

Good point. Some rich people might be content with having an AI toy that doesn't do much useful stuff, though.

[Ares Land]

The sort of AI you're envisioning basically means post-scarcity. The question of capitalism vs. socialism would be moot -- there'd no longer be any economics, really.

I'm still not convinced we're anywhere near that sort of AI.

[Torco]

You don't need sentience or sapience to get work done. You can potentially automate your entire workforce with non-sentient AIs. Plus, non-sentient AIs won't ask for a salary.

hopefully!

But did you apodictically deduce that you didn't apodictically deduce anything else?

I just took a tally of all the times people go "see? this is deductively true" while saying things which are obviously false, also, after decades of living as a human, realized I hadn't used deduction in my regular life even once, outside of reading Plato or playing around with Thales' theorem in class... so yes .

ai consumers when

when AI workers. no earlier.

Proposed Axiom of Impossibility of Artificial Intelligence

I think the problem with this is the same problem with the notion of post-scarcity: it assumes that work will/can run out. This is understandable, as within capitalism "work" kind of means "things someone is willing to pay a particular person at least enough so that they can barely stave off starvartion" and, in that sense, there is a finite amount of work. but in reality, outside the ideological assumptions of capitalism, work (in the sense of things which would be good, advantageous, comfortable or profitable for someone if they were done) is kind of infinite: and therefore, there is a relatively large supply of tasks that it would be good if a person/thing/ai were able to do forever for almost no money. autocorrect could be done by poor filipinos working in vast datacenters, but it's easier and thus more profitable for my OEM if my phone does it itself.

[chris_notts]

Incidentally, I tested GPT3 and GPT4 as a way to get feedback on conlang grammars with mixed results. GPT3 struggles to say anything useful, and starts either answering questions other than the one I asked or hallucinating fake examples and grammar for natural languages...

GPT-4 did some of that, but was much better at giving useful feedback and answering the question I was trying to ask, and not making stuff up.

Even GPT-4 was not really amazing though. It generates academic sounding waffle really well, but doesn't really offer that many deep thoughts.

[Raphael]

but doesn't really offer that many deep thoughts.

Then again, neither do a lot of human beings.

[Raphael]

I am getting a bit worried what GPT-5 might do if it turns out to be as good compared to GPT-4 as GPT-4 is compared to GPT-3, though.

[chris_notts]

but doesn't really offer that many deep thoughts.

Then again, neither do a lot of human beings.

True, most humans are also shockingly bad at critiquing conlang grammars.

[rotting bones]

Proposed Axiom of Impossibility of Artificial Intelligence

1) As long as there are Filipino teenagers willing to work for $5 a day, that will always be easier than AI. You get turing-complete intelligence, and someone else eats the manufacturing and R&D costs.

2) When there are no more Filipino teenagers willing to work for $5 a day, there is no longer any market for AI. Our global economy cannot function without income disparity, not to mention all the supply chains that would stop dead without cheap labor.

In other words, we can only want Turing-complete robots if we have leisure time to sit around and make sexy chat with anime girls while the global south peels our grapes for us. But the existence of the global south allows us to outsource all the troublesome work that AIs would do.

This is not the same as the "industrialization trap" people talk about with, say, early modern China. That was an avoidable dilemma, as seen in Europe, which managed to increase both automated productivity and wages through the clever strategy of going to other people's countries and shooting at them. That won't work here, because the fact that Filipino teenagers are poor is not a bug, but a feature. We depend on them having less money than us, in order to make the flow of capital and goods function.

As the glorious Knight of Espionage pointed out, AI could be used to automate production.

Marxian economics distinguishes between use value and exchange value to avoid this confusion. The products of labor have uses apart from market exchange. My proposal to replace money with votes is based precisely on the difference between exchange demand and use demand.

The need is simply for an imbalance. Aviva Chomsky, the only good Chomsky, wrote extensively about this and the migration of the textile industry.

I like Noam Chomsky. I'm a very long time fan of his. I know he's often wrong. He has always been often wrong, but we love him in theoretical CS. Also, he popularizes facts about capitalism and imperialism that are still not common knowledge outside leftist bubbles.