Hacker News

It isn't possible to have "just probabilistic" (maybe a philosophical exception could be made for a uniform random distribution or whatever provides the little dose of randomness required to get nondeterministic results). Probabilities are always in context of a model. LLMs model language but language itself is a model of something else. My money would have been on language modelling nonsense, but that is quite clearly not the case. Turns out it models the world and so do LLMs.

The model is the thing which is learned in order to make the probabilistic prediction with low entropy.

The literal definition of a model is "an informative representation of an object, person, or system". I think you mean something else though, what are you trying to express exactly?

Nothing about an LLM is “just”. In what precise sense do you mean it is probabilistic?

Maybe, but a claim about what and LLM is not is still a claim about what it can or cannot do. And specifically:

> without any reference to meaning

is vague, but I read it as actually quite a strong claim about the limitations of LLMs. I don’t think it would be possible for LLMs to do long chains of correct mathematical reasoning about novel problems that they haven’t seen before “without any reference to meaning.” That simply isn’t possible just by regurgitating and remixing random chunks of training data. Therefore I consider the stochastic parrots picture of LLMs to be wrong.

It might have been an accurate picture in 2020. It is not an accurate picture now. What is often missed in these discussions is that LLM training now looks totally different than it did a couple years ago. RLVR completely changed the game, allowing LLMs to actually do math and code well, among other things.

> She says explicitly it's not an empirical hypothesis. It's just a label for how they function.

Then… what’s the point of the label, if it’s not making any empirically-meaningful claims about LLMs at all? I know that LLMs involve sampling over a distribution of output logits. I’ve written code to do it. So what? I know they have statistical elements. Yet I don’t go around calling LLMs stochastic parrots, because that label implies a whole lot of claims about LLMs that I don’t think are true any longer, like that they are just regurgitating and remixing training data and can’t successfully model structured systems (like mathematics or programming).

Yes, she addresses this by denying that she's made any empirical hypothesis, but in a way that's some combination of disingenuous and confused.

She also says:

> What I am trying to do... is to help people understand what these systems actually are

Can a phrase that has no empirical content aid people in understanding an empirical phenomenon?

> the astonishing willingness of so many to... turn to synthetic text... for all kinds of weighty decisions.

Why is this astonishing, if the nature of these models as "stochastic parrots" places no limitations whatosever on their empirical capabilities, reliability, etc?

> the field of linguistics is particularly relevant in this moment, as a linguist’s eye view on language technology is desperately needed to help make wise decisions about how we do and don’t use these products

Is it wise to make decisions about a product on the basis of information that has no relevance to how it is actually likely to behave?

(It may be, if one has ethical concerns with "data theft, the exploitative labor practices", etc -- but one could have such concerns about any kind of product, not just a "stochastic parrot", and linguists are not generally academia's experts on, e.g., labor practices.)

Is your justification in dismissing Fields medalists that they are impressed by funding? Not even receiving it (I assume you say this because Tao is not funded by AI for Math, but rather an advisor for it)?

Not only would it be a leap to suggest that people automatically lose their integrity by taking funds for projects they believe are useful, especially after involvement with adjacent fields, but you are suggesting merely being impressed by a fund is enough to dismiss their views?

You also have no evidence that Renaissance Philanthropies is a front for VC companies. All news coverage indicates that they seek to be an alternative for high net worth individuals engaging in philanthropy.

Many people discovering Erdos results, engaging in Olympiads etc, are doing so with publicly available models and publish the resources used in the process.

Automated theorem provers are not new, in fact they are very old. One of the most automated is ACL2, which uses the well studied waterfall method (unrelated to waterfall development).

LLMs certainly use something similar, except they understand text as input. LLMs, especially used for marketing stunts, have way more computing power available than any theorem prover ever had. They probably do random restarts if a proof fails which amounts to partially brute forcing.

Lawrence Paulson correctly complained about some of the hype that Lean/LLMs are getting.

ACL2 even uses formulaic text output that describes the proof in human language, despite being all in Common Lisp and not a mythical clanker.

They do not think and use old and well established algorithms or perhaps novel ones that were added.

And also create novel math proofs.