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

GPT-5 is behind schedule

wsj.com

Post reply on HN
Print view View original

301–310 of 1001 posts

Re: GPT-5 is behind schedule

#301

Post by overgard »

simonw wrote: post #113

"OpenAI’s is called GPT-4, the fourth LLM the company has developed since its 2015 founding." - that sentence doesn't fill me with confidence in the quality of the rest of the article, sadly.

The article definitely has issues, but to me what's relevant is where it's published. The smart money and experts without a vested interest have been well aware LLMs are an expensive dead for over a year and have been saying as much (Gary Marcus for instance). That this is starting to enter mainstream consciousness is what's newsworthy.

Re: GPT-5 is behind schedule

#302

Post by dartos »

nialv7 wrote: post #101
dartos wrote: post #50

Earlier quoted context omitted.

I’m curious how, if at all, the plan to get around compounding bias in synthetic data generated by models trained in synthetic data.

synthetic data is fine if you can ground the model somehow. that's why the o1/o3's improvements are mostly in reasoning, maths, etc., because you can easily tell if the data is wrong or not.

That makes a lot of sense.

Binary success criteria has very little room for bias.

Re: GPT-5 is behind schedule

#303

Post by jjcm »

The results of this article are going to be fascinating. Realistically, WSJ has a far wider audience than the tech echo chamber, and the general public is only aware of GPT, not o1/o3.

Outsiders will likely read this article and think, “AI is running out of steam”, because GPT-5 is behind.

Those closer to this know of the huge advancements o3 just made yesterday, and will have a complete opposite conclusion.

It will be interesting to see people’s take away from this. I think WSJ missed the mark here with the headline and the takeaway their audience will get from the article.

Re: GPT-5 is behind schedule

#304

Post by radioactivist »

justatdotin wrote: post #297
radioactivist wrote: post #67

Earlier quoted context omitted.

I had a similar experience with regular o1 about integral that was divergent. It was adamant that it wasn't and would respond to any attempt at persuasion with variants of "its a standard integral" with a "subtle cancellation". When I asked for any source for this standard integral it produced references to support its argument that existed but didn't actually contain the integral. When I told it the references didn'…

I've also had it invent non-existent references. > being this confidently wrong (and "lying" when confronted with it) is troubling. I don't find it troubling. I like being reminded to distrust and confirm everything it offers.

The troubling part is that the references themselves existed -- one was an obscure Russian text that is difficult to find (but is exactly where you'd expect to find this kind of result, if it existed).

Re: GPT-5 is behind schedule

#305

Post by LarsDu88 »

missedthecue wrote: post #271
dkobia wrote: post #232

Earlier quoted context omitted.

AGI is the Sisyphean task of our age. We’ll push this boulder up the mountain because we have to, even if it kills us.

Do we know LLMs are the path to AGI? If they're not, we'll just end up with some neat but eye wateringly expensive LLMs.

The autoregressive transformer LLMs aren't even the only way to do text generation. There are now diffusion based LLMs, StripedHyena based LLMs, and float matching based LLMs.

There's a wide amount of research into other sorts of architectures.

Re: GPT-5 is behind schedule

#306

Post by famouswaffles »

mrbungie wrote: post #236
famouswaffles wrote: post #167

Earlier quoted context omitted.

Not really. o3-low compute still stomps the benchmarks and isn't anywhere that expensive and o3-mini seems better than o1 while being cheaper. Combine that with the fact that LLM inference has reduced orders of magnitudes in cost the last few years and hampering over the inference costs of a new release seems a bit silly.

It is still not economical: in Arc at least 20 usd for task vs ~3 usd for a human (avg mturker) for the same perf.

Not necessarily. And this is the problem with ARC that people seem to forget.

- It's just a suite of visual puzzles. It's not like say GSM8K where proficiency in it gives some indication on Math proficiency in general.

- It's specifically a suite of puzzles that LLMs have shown particular difficulty in.

Basically how much compute it takes to handle a task in this benchmark does not correlate with how much it will take LLMs to compute tasks that people actually want to use LLMs for.

Re: GPT-5 is behind schedule

#307

Post by coldtea »

Eisenstein wrote: post #298
coldtea wrote: post #267

Earlier quoted context omitted.

> What do you think the last few years have been all about? Next token language-based predictors with no more intelligence than brute force GIGO which parrot existing human intelligence captured as text/audio and fed in the form of input data. 4o agrees: "What you are describing is a language model or next-token predictor that operates solely as a computational system without inherent intelligence or understanding. T…

You have described something but you haven't explained why the description of the thing defines its capability. This is a tautology, or possibly a begging of the question, which takes as true the premise of something (that token based language predictors cannot be intelligent) and then uses that premise to prove an unproven point (that language models cannot achieve intelligence). You did nothing at all to demonstrat…

>You did nothing at all to demonstrate why you cannot produce an intelligent system from a next token language based predictor

Sorry, but the burden of proof is on your side...

The intelligence is in the corpus the LLM was fed with. Using statistics to pick from it and re-arrange it gives new intelligent results because the information was already produced by intelligent beings.

If somebody gives you an excerpt of a book, it doesn't mean they have the intelligence of the author - even if you have taught them a mechanical statistical method to give back a section matching a query you make.

Kids learn to speak and understand language at 3-4 years old (among tons of other concepts), and can reason by themselves in a few years with less than 1 billionth the input...

>What GPT says about this is completely irrelevant.

On the contrary, it's using its very real intelligence, about to reach singularity any time now, and this is its verdict!

Why would you say it's irrelevant? That would be as if it merely statistically parroted combinations of its training data unconnected to any reasoning (except of that the human creators of the data used to create them) or objective reality...

Re: GPT-5 is behind schedule

#308

Post by overgard »

maxrmk wrote: post #276
simonw wrote: post #113

"OpenAI’s is called GPT-4, the fourth LLM the company has developed since its 2015 founding." - that sentence doesn't fill me with confidence in the quality of the rest of the article, sadly.

I was wondering about this one too... > At best, they say, Orion performs better than OpenAI’s current offerings, but hasn’t advanced enough to justify the enormous cost of keeping the new model running. wdym "keep it running"?

Well, those server farms don't pay for themselves.

Re: GPT-5 is behind schedule

#309

Post by unshavedyak »

pbhjpbhj wrote: post #296
unshavedyak wrote: post #283

Earlier quoted context omitted.

> which is priceless This also isn't true. It'll clearly have a price to run. Even if it's very intelligent, if the price to run it is too high it'll just be a 24/7 intelligent person that few can afford to talk to. No?

Computers will be the size of data centres, they'll be so expensive we'll queue up jobs to run on them days in advance, each taking our turn... history echoes into the future...

Yea, and those statements were true. For a time. If you want to say "AGI will be priceless some unknown time into the future" then i'd be on board lol. But to imply it'll be immediately priceless? As in no cost spent today wouldn't be immediately rewarded once AGI exists? Nonsense.

Maybe if it was _extremely_ intelligent and it's ROI would be all the drugs it would instantly discover or w/e. But lets not imply that General Intelligence requires infinitely knowing.

So at best we're talking about an AI that is likely close to human level intelligence. Which is cool, because we have 7+ billion of those things.

This isn't an argument against it. Just to say that AGI isn't "priceless" in the implementation we'd likely see out of the gate.

Re: GPT-5 is behind schedule

#310

Post by School-Cotton »

SahAssar wrote: post #299
xanderlewis wrote: post #206

Earlier quoted context omitted.

It’s somehow funny to hear a British company being described as ‘in Europe’, but I suppose you’re technically correct…

The UK is part of Europe. It's technically, geographically, politically, historically, lingustially, tectonically and socially correct. In what ways is it not?

Are Cuba or Haiti part of North America? A lot of British people feel like their civilization is meaningfully distinct from “Europe”, even though they’re part of it in a technical geographical sense.
Post reply on HN