Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

reuters.com

Post reply on HN
Print view View original

641–650 of 1001 posts

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#641

Post by uoaei »

wbhart wrote: post #565
neilk wrote: post #532

Earlier quoted context omitted.

I am neither a mathematician or LLM creator but I do know how to evaluate interesting tech claims. The absolute best case scenario for a new technology is that it when it seems like a toy for nerds, and doesn't outperform anything we have today, but the scaling path is clear. Its problems just won't matter if it does that one thing with scaling. The web is a pretty good hypermedia platform, but a disastrously bad pla…

How on earth could you evaluate the scaling path with too little information. That's my point. You can't possibly know that a technology can solve a given kind of problem if it can only so far solve a completely different kind of problem which is largely unrelated! Saying that performance on grade-school problems is predictive of performance on complex reasoning tasks (including theorem proving) is like saying that a…

Any claims of objective, quantitative measurements of "scaling" in LLMs is voodoo snake oil when measured against some benchmarks consisting of "which questions does it answer correctly". Any machine learning PhD will admit this, albeit only in a quiet corner of a noisy bar after a few more drinks than is advisable when they're earning money from companies who claim scaling wins on such benchmarks.

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#642

Post by starbugs »

red75prime wrote: post #418
RcouF1uZ4gsC wrote: post #344

This almost feels like the confirmation bias that some religious people have where they see a “miracle” in everything. These AI researchers have bought into the belief that superhuman AGI is right around the corner. Thus, they will interpret everything in light of that. This also brings to mind the story of the Googler who was convinced that the internal Google AI had come alive. However, Bard doesn’t give the same v…

Ugh. We have a working example of a physical system that implements intelligence (the brain) in contrast to no evidence of all-powerful dude in the sky. Why these analogies keep popping up? How can you know that AGI is not around the corner? Compute available to the corporations is already in a ballpark of some estimates of the brain's computational capacity. What's left is unknown unknows. And the researches working…

Judging from the quality of GPT4 output that I get, AGI is not around the corner for a long time.

This whole thing seems like extreme overhype squared carried out in a very unfortunate public soap opera setting.

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#643

Post by neta1337 »

Smith42 wrote: post #184
jajabinks11 wrote: post #181

Let's repeat this: current LLMs != AGI. It will never be. It could be used for knowledge store, retrieval, synthesis, and low lever reasoning. There is at least a few decades worth of work remaining for it to reach the level of AGI. The proponents in this AI gold rush are casually throwing around the term without any real thought.

Wishful thinkin buddy

Buy more shovels buddy

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#644

Post by curling_grad »

nostrademons wrote: post #593
wbhart wrote: post #491

I feel very comfortable saying, as a mathematician, that the ability to solve grade school maths problems would not be at all a predictor of ability to solve real mathematical problems at a research level. The reason LLMs fail at solving mathematical problems is because: 1) they are terrible at arithmetic, 2) they are terrible at algebra, but most importantly, 3) they are terrible at complex reasoning (more specifica…

What I wonder, as a computer scientist: If you want to solve grade school math problems, why not use an 'add' instruction? It's been around since the 50s, runs a billion times faster than an LLM, every assembly-language programmer knows how to use it, every high-level language has a one-token equivalent, and doesn't hallucinate answers (other than integer overflow). We also know how to solve complex reasoning chains…

Actually, OpenAI did a research[0] on solving some hard math problems by integrating language model and Lean theorem prover some time ago.

[0]: https://openai.com/research/formal-math

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#645

Post by Closi »

nostrademons wrote: post #593
wbhart wrote: post #491

I feel very comfortable saying, as a mathematician, that the ability to solve grade school maths problems would not be at all a predictor of ability to solve real mathematical problems at a research level. The reason LLMs fail at solving mathematical problems is because: 1) they are terrible at arithmetic, 2) they are terrible at algebra, but most importantly, 3) they are terrible at complex reasoning (more specifica…

What I wonder, as a computer scientist: If you want to solve grade school math problems, why not use an 'add' instruction? It's been around since the 50s, runs a billion times faster than an LLM, every assembly-language programmer knows how to use it, every high-level language has a one-token equivalent, and doesn't hallucinate answers (other than integer overflow). We also know how to solve complex reasoning chains…

Why would we teach kids maths then, when they can use a calculator? It's much easier and faster for them.

I believe it's because having a foundational understanding of maths and logic is important when solving other problems, and if you are looking to create an AI that can generally solve all problems it should probably have some intuitive understanding of maths too.

i.e. if we want an LLM to be able to solve unsolved theorems in the future, this requires a level of understanding of maths that is more than 'teach it to use a calculator'.

More broadly, I can imagine a world where LLM training is a bit more 'interactive' - right now if you ask it to play a game of chess with you it fails, but it has only ever read about chess and past games and guesses the next token based on that. What if it could actually play a game of chess - would it get a deeper appreciation for the game? How would this change it's internal model for other questions (e.g. would this make it answer better at questions about other games, or even game theory?)

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#646

Post by yinser »

casualscience wrote: post #314

I don't really know what kind of breakthrough they could achieve. The only other step function improvements I could imagine right now are: 1. A great technique for memory banking: e.g. A model which can have arbitrarily large context windows (i.e. like a human who remembers things over long periods of time). 2. Better planning abilities: e.g. A model which can break problems down repeatedly with extremely high succes…

Complete speculation but they don’t have to be referring, if there is anything at all, to heuristic search and dynamic programming. No LLM involvement whatsoever.

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#647

Post by panarky »

nostrademons wrote: post #593
wbhart wrote: post #491

I feel very comfortable saying, as a mathematician, that the ability to solve grade school maths problems would not be at all a predictor of ability to solve real mathematical problems at a research level. The reason LLMs fail at solving mathematical problems is because: 1) they are terrible at arithmetic, 2) they are terrible at algebra, but most importantly, 3) they are terrible at complex reasoning (more specifica…

What I wonder, as a computer scientist: If you want to solve grade school math problems, why not use an 'add' instruction? It's been around since the 50s, runs a billion times faster than an LLM, every assembly-language programmer knows how to use it, every high-level language has a one-token equivalent, and doesn't hallucinate answers (other than integer overflow). We also know how to solve complex reasoning chains…

>> the ability to solve grade school maths problems would not be at all a predictor of ability to solve real mathematical problems at a research level

> If you want to solve grade school math problems, why not use an 'add' instruction?

Certainly the objective is not for the AI to do research-level mathematics.

It's not really even to do grade-school math.

The point is that grade-school math requires reasoning capability that transcends probabilistic completion of the next token in a sequence.

And if Q-Star has that reasoning capability, then it's another step-function leap toward AGI.

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#648

Post by riku_iki »

adastra22 wrote: post #515
wbhart wrote: post #491

I feel very comfortable saying, as a mathematician, that the ability to solve grade school maths problems would not be at all a predictor of ability to solve real mathematical problems at a research level. The reason LLMs fail at solving mathematical problems is because: 1) they are terrible at arithmetic, 2) they are terrible at algebra, but most importantly, 3) they are terrible at complex reasoning (more specifica…

Back-tracking is a very nearly solved problem in the context of Prolog-like languages or mathematical theorem provers (as you probably well know). There are many ways you could integrate an LLM-like system into a tactic-based theorem prover without having to restart from the beginning for each alternative. Simply checkpointing and backtracking to a checkpoint would naively improve upon your described Monte Carlo algo…

> That's just what I came up with after thinking on it for 2 minutes. I'm sure they have even better ideas.

the thing is that ideas not necessary easy to implement. There will be many obstacles on route you described:

- quality of provers, is there good ergo provers which also can run at large scales (say billions of facts)

- you need some formalization approach, probably LLM will do some work, but we don't know what will be quality

- LLM likely will generate many individual factoids, which are losely compatible, contradicting, etc, and untrivial effort is required to reconcile and connect them

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#649

Post by xcv123 »

Terr_ wrote: post #638
gmerc wrote: post #600

Earlier quoted context omitted.

After a year the entire argument you make boils down to “so far”.

Whereas your post sounds like "Just give the approach more time, it shall continue to incrementally improve until it finally works someday, cuz reasons. " Early attempts at human flight approached it by strapping wings to people's arms and flapping: Do you think that would have eventually worked too, if only we had just given it a bit more time and faith?

> Just give the approach more time, it shall continue to incrementally improve until it finally works someday, cuz reasons

Yes. Because we haven't yet reached the limit of deep learning models. GPT-3.5 has 175 billion parameters. GPT-4 has an estimated 1.8 trillion parameters. That was nearly a year ago. Wait until you see what's next.

Re: OpenAI researchers warned board of AI breakthrough ahead of CEO ouster

#650

Post by riwsky »

casualscience wrote: post #314

I don't really know what kind of breakthrough they could achieve. The only other step function improvements I could imagine right now are: 1. A great technique for memory banking: e.g. A model which can have arbitrarily large context windows (i.e. like a human who remembers things over long periods of time). 2. Better planning abilities: e.g. A model which can break problems down repeatedly with extremely high succes…

Nah dude they just need to give baby the iPad and let it watch literally all of YouTube
Post reply on HN