Will scaling work? • Hacker News Board

Re: Will scaling work?

#241

Post by throwup238 » Thu, Dec 28, 2023, 2:10 AM UTC

zoogeny wrote: post #147

I was thinking last night about LLMs with respect to Wittgenstein after watching this interesting discussion of his philosophy by John Searle [1]. I think Wittgenstein's ideas are pertinent to the discussion of the relation of language to intelligence (or reasoning in general). I don't meant this in a technical sense (I recall Chomsky mentioning that almost no ideas from Wittgenstein actually have a place in modern l…

Wittgenstein is the perfect lens through which to be skeptical about LLMs and AGI and I think you may be the one not fully engaging with his work. He saw that languages are inseparable from the context in which they are used and that context is much bigger than language itself. Part of learning those language games is experimenting in the real world - interacting with other people, playing language games, and seeing how they react is how we build up our internal dictionaries and innate knowledge. The fidelity of text is simply too low to communicate the amount of information humans use to build up general intelligence.

Without the ability to interact with the physical world, LLMs will never be able to reach AGI. It can kind of simulate it to some extent, but it'll never get there

LLMs can't even form their own memories, the context has to be explicitly fed back to them.

Re: Will scaling work?

#242

Post by lnxg33k1 » Thu, Dec 28, 2023, 2:36 AM UTC

qudat wrote: post #235

lnxg33k1 wrote: post #183

Earlier quoted context omitted.

Imagine telling those same people in the 50s that all those changes in productivity would come for the benefit of no one since the work week would be the same and purchasing power would decline

Such a wild take. Would you want to live in the 50s? I definitely would not.

Well, I'd at least be able to buy a house, and sustain a family with a single job

Re: Will scaling work?

#243

Post by zoogeny » Thu, Dec 28, 2023, 2:58 AM UTC

throwup238 wrote: post #241

zoogeny wrote: post #147

I was thinking last night about LLMs with respect to Wittgenstein after watching this interesting discussion of his philosophy by John Searle [1]. I think Wittgenstein's ideas are pertinent to the discussion of the relation of language to intelligence (or reasoning in general). I don't meant this in a technical sense (I recall Chomsky mentioning that almost no ideas from Wittgenstein actually have a place in modern l…

Wittgenstein is the perfect lens through which to be skeptical about LLMs and AGI and I think you may be the one not fully engaging with his work. He saw that languages are inseparable from the context in which they are used and that context is much bigger than language itself. Part of learning those language games is experimenting in the real world - interacting with other people, playing language games, and seeing…

> He saw that languages are inseparable from the context in which they are used

That is one of the things that stood out to me in Searle's summary of his later work because I consider how the transformer architecture works and the way in which the surrounding context plays into the meaning of the words.

> Part of learning those language games is experimenting in the real world

It is interesting that the article we are responding to talks about how we have only just begun to experiment with RL on top of transformers. In the same way that Alpha Go engaged in adversarial play we can envision LLMs being augmented to play language games amongst themselves. That may result in their own language, distinct from human language. But it also may result in the formation of intelligence surpassing human intelligence.

> The fidelity of text is simply too low to communicate the amount of information humans use to build up general intelligence.

This does not at all follow from anything I've encountered in Wittgenstein. It is an empirical claim that we (as in humanity) are going to test and not something that I would argue either one of us can know simply reasoning from first principles.

What does follow for me is closer to what Steven Pinker has been proposing in his own critiques of LLMs and AGI, which is that there is no necessary correlation between goal seeking (or morality) and intelligence. I also feel this is concordant with Wittgenstein's own work.

> Without the ability to interact with the physical world, LLMs will never be able to reach AGI

Again, a confident claim that is based on nothing other than your own belief. As I stated in my last comment, I am excited to see if that is empirically true or false. We are definitely going to scale up LLMs in the coming decade and so we are likely to find out.

My suspicion is that people don't want this scaling up to work because it would force them to let go of metaphysical commitments they have on both the nature of intelligence as well as the nature of reality. And for this reason they are adamantly disbelieving in even the possibility before the evidence has been gathered.

I'm happy to stay agnostic until the evidence is in. Thankfully, it shouldn't take too long so I may be lucky enough to find out in my own lifetime.

Re: Will scaling work?

#244

Post by goatlover » Thu, Dec 28, 2023, 3:28 AM UTC

qudat wrote: post #235

lnxg33k1 wrote: post #183

Earlier quoted context omitted.

Imagine telling those same people in the 50s that all those changes in productivity would come for the benefit of no one since the work week would be the same and purchasing power would decline

Such a wild take. Would you want to live in the 50s? I definitely would not.

It doesn't need to be all or nothing. It's possible to acknowledge that the 50s had a better economic outlook for the middle class in developed countries than it does now, despite the incredible advances in computing. And you might still prefer to live now, for social reasons or because you prefer the computing advances, despite it not delivering economically as much.

Either way, it does highlight a serious economic concern as computing continues to advance. The majority of the wealth is increasingly being concentrated by our tech overlords.

Re: Will scaling work?

#245

Post by throwup238 » Thu, Dec 28, 2023, 3:30 AM UTC

zoogeny wrote: post #243

throwup238 wrote: post #241

Earlier quoted context omitted.

Wittgenstein is the perfect lens through which to be skeptical about LLMs and AGI and I think you may be the one not fully engaging with his work. He saw that languages are inseparable from the context in which they are used and that context is much bigger than language itself. Part of learning those language games is experimenting in the real world - interacting with other people, playing language games, and seeing…

> He saw that languages are inseparable from the context in which they are used That is one of the things that stood out to me in Searle's summary of his later work because I consider how the transformer architecture works and the way in which the surrounding context plays into the meaning of the words. > Part of learning those language games is experimenting in the real world It is interesting that the article we ar…

It's not a belief or logically derived claim, it's as close to empirical fact as we're going to get. We have zero evidence that intelligence without physical experimentation is possible because we have no other examples of intelligence except humans (and nonhuman animals), all of whom learned experimentally with a physical feedback loop. Even the most extreme cases like Helen Keller depended on it - her story is perhaps far more useful to grounding theories about AGI than any philosophical text as Wittgenstein himself would likely argue (Water!). His contempt, for lack of a better word, for philosophy on those terms is clear.

I'm excited to see how LLMs scale but it won't reach AGI without a much richer architecture that is capable of experimentation, capable of playing "language games" with other humans and remembering what it learned.

(I'm fairly certain of my views given my experience in neuroscience but it's fun to talk Wittgenstein in the context of LLMs, something that's been conspicuously missing. Sadly I don't believe discussions of AGI are fruitful, just what LLMs can teach us about the nature of language)

Re: Will scaling work?

#246

Post by goatlover » Thu, Dec 28, 2023, 3:32 AM UTC

Xelynega wrote: post #224

lnxg33k1 wrote: post #183

Earlier quoted context omitted.

Imagine telling those same people in the 50s that all those changes in productivity would come for the benefit of no one since the work week would be the same and purchasing power would decline

Don't have to imagine, same thing is happening right now with LLMs. I see "AI safety" brought up as a laughable attempt at stopping the progress of LLMs, when in reality the people talking about "AI safety" are the people trying to say that the majority will not benefit from this technology.

I think the AI safety people are saying that whatever benefits LLMs bring to the masses might be outweighed by the costs. We've seen this with social media. And on the extreme end, there's the existential concern. If we do ever get to AGI, all bets are off from where we stand right now, because nobody has the faintest clue how a human-level (or beyond) intelligence will play out in society.

Re: Will scaling work?

#247

Post by skippyboxedhero » Thu, Dec 28, 2023, 3:44 AM UTC

joe_the_user wrote: post #172

skippyboxedhero wrote: post #116

Earlier quoted context omitted.

His view of the Industrial Revolution is completely wrong. Societies pre-IR had multiple periods where energy usage increased significantly, some of them based specifically around coal. No IR. Early IR was largely based around the usage of water power, not coal. IR was pure innovation, people being able to imagine and create the impossible, it was going straight to nuclear already. Ironically, someone who is an innov…

Societies pre-IR had multiple periods where energy usage increased significantly, some of them based specifically around coal. No IR. That's a straight up misstatement of the parent argument - the parent argued that coal was necessary, not that coal sufficient. True or not, the argument isn't refuted by the IR starting with water power either. And pairing this with "anti-woke" jabs is discourse-diminishing stuff. The…

It isn't a misstatement, it is (as I explained) a common argument that falls within Marxist/materialist viewpoints such as Robert Allen's account of global IR, it was the dominant viewpoint when this guy was at uni, you often hear people in the UK of that age saying this stuff (I know, I studied economic history during this time) but the field has continued to progress since then. Water energy wasn't a "starting", it was the IR. Coal didn't come until significantly later (and the major issue with materialist history is also what happened in the 20th century, you had multiple countries attempt and fail to industrialise using this idea that energy intensity was the only thing that mattered, the biggest issue with anti-Eurocentric theory is that it was designed to explain a 40-year period around the end of the 18th century and completely fails to generalize).

What is anti-woke? You realise that stuff existed before zoomers starting saying everything was woke/anti-woke. Eurocentrism is a school of thought within economic history, it is nothing to do with wokeism...I have no idea how these two things are related apart from you trying to relate it to something you understand, i.e. pop culture.

"Pure innovation" fluff is the dominant theory today, McCloskey's books are the most important ones in this school. To call this "fluff" suggests ignorance rather than the superiority that you seem to be trying to portray.

Petroleum wasn't a key ingredient of IR...at this point, I am assuming you know nothing about basic aspects of economic history because petroleum wasn't widely used as a fuel until the 1930/40s (again, you seem intent on talking about things that you know rather than the actual subject).

Re: Will scaling work?

#248

Post by goatlover » Thu, Dec 28, 2023, 3:54 AM UTC

majkinetor wrote: post #160

ketzo wrote: post #152

Earlier quoted context omitted.

I mean, very broad strokes, but I can see GP’s point. - people eat plants and animals - people pay money for goods and services - there are countries, sometimes they fight, sometimes they work together - men and women come together to create children, and often raise those children together etc, etc, etc The “bones” of what make up a capital-S Society are pretty much the same. None of these things had to stay the sam…

VERY broad strokes. We also still have a Sun, and the stars. Internet and the last 30 years tech did change things dramatically. I bet that most people would feel handicapped if they were teleported just 50 years back. We got into this type of life progressively, so people didn't notice the change, even though it was dramatic. The same phenomena with gradient changes happen on physiological level too, this is not dif…

It's dramatic, but there are plenty of things in society that are similar to what they were 50 years ago. It's not like people from 50 years ago would be incapable of understanding those changes if you explained them. Which is a bit different than 500 years ago.

At least if we're using the technological singularity was what constitutes fundamental societal change in unpredictable ways. The singularity people think AGI is going to fundamentally change everything, even more than what the past 500 years has done. Certainly many magnitudes of order more than the last 50 years. And they think it will happen much faster.

Re: Will scaling work?

#249

Post by goatlover » Thu, Dec 28, 2023, 4:07 AM UTC

stevenhuang wrote: post #208

timacles wrote: post #89

Earlier quoted context omitted.

> It is not an unreasonable hypothesis that brains evolved to solve a similar sequence modeling problem. The real world is random, requires making decisions on incomplete information in situations that have never happened before. The real world is not a sequence of tokens. Consciousness requires instincts in order to prioritize the endless streams of information. One thing people dont want to accept about any AI is t…

The real world is informational. If the world is truly random and devoid of information, you wouldn't exist.

Information is a loaded word. Sure, you can say that based on our physical theories, you can think of the world that way, but information is what's meaningful to us amongst all the noise of the world. Meaningful for goals like survival and reproduction from our ancestors. Nervous systems evolved to help animals decide what's important to focus on. It's not a premade data set, the brain makes it meaningful in context of it's environment.

Re: Will scaling work?

#250

Post by goatlover » Thu, Dec 28, 2023, 4:09 AM UTC

measured_step wrote: post #101

timacles wrote: post #89

Earlier quoted context omitted.

> It is not an unreasonable hypothesis that brains evolved to solve a similar sequence modeling problem. The real world is random, requires making decisions on incomplete information in situations that have never happened before. The real world is not a sequence of tokens. Consciousness requires instincts in order to prioritize the endless streams of information. One thing people dont want to accept about any AI is t…

How do our base reptilian brains reason? We don't know the specifics, but unless it's magic, then it's determined by some kind of logic. I doubt that logic is so unique that it can't eventually be reproduced in computers.

Reptiles didn't use language tokens, that's for sure. We don't have reptilian brains anyway, it's just that part of our brain architecture evolved from a common ancestor. The stuff that might function somewhat similar to an LLM is most likely in the neocortex. But that's for neuroscientists to figure out, not computer scientists. Whatever the case is, it had to have evolved. LLMs are intelligently designed by us, so we should be a little cautious in making that analogy.