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

Why I'm still bearish on LLMs after Navier-Stokes

dank.systems

Post reply on HN
Print view View original

281–290 of 642 posts

Re: Why I'm still bearish on LLMs after Navier-Stokes

#281

Post by zahlman »

joefourier wrote: post #31
carodgers wrote: post #17

This April 2026 paper is a fun and related read. https://arxiv.org/html/2509.24239v4 Researchers asked frontier models to play chess. Have a look at the MAR rates in Table 3. When not explicitly told which moves were legal, no model identified legal moves at a rate better than 80%. Many asked for more illegal moves than legal moves. And even when explicitly told which moves were legal, the models continued to ask for…

> current frontier models > Gemini 2.5 Pro, O3, Claude Sonnet 3.7 and ChatGPT 4.1 The gap in capabilities between those models which they tested, and actual current frontier ones is enormous. I would not trust that any conclusions they made are applicable.

Just now I tried prompting logged-out ChatGPT (which at least claims to be 5.6-Luna) with:

> Let's play a game of chess. You can take White. Please draw an ASCII rendition of the board after each move, so that we can be clear about the position.

(I hoped the latter requirement would help it be "not blindfolded"; last time I used a Lichess demo board to track the position in another tab, because I have no talent for blindfold chess.)

For the first couple of moves it redrew the board after each move; then it started only drawing it after my moves. And on moves 5 and 6 it dropped two minor pieces for pawns in a row without any meaningful positional advantage, and after I captured the second time, it redrew a board that was simply missing one of my pieces for no reason.

It actually played better when not prompted to draw a board; in the previous session, it was spontaneously giving running commentary, which I assume was based off all the "book" theory in its training data, but it still completely fell apart at early midgame.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#282

Post by tsimionescu »

Gregkion wrote: post #246
minraws wrote: post #81

Earlier quoted context omitted.

So AGI needs to be trained on something to work well on it. Lovely reasoning we have right here. Delusion runs deep in HN circles. I say that as someone heavily invested in AI startups and projects and as someone working in the field. I think most people on HN should touch grass and find real human contact. Lmao Incredible reasoning all around here.

An AGI doesn't stand for 'perfect intelligence' it stands for artificial general intelligence. And no an AGI system doesn't need to play chess on a certain level to be disruptive to you and me and whole industries. It only needs to be as good as a person and cheaper. Just because you define AGI as something it doesn't has to be,doesn't mean i need to touch grass. This chess comparision is one of the most ignorant and…

Do you know what the "General" in "Artificial General Intelligence" means? It specifically means that the AGI adapts to novel domains that it hasn't been trained on - its training generalizes to real world problems.

That doesn't mean it has to be extraordinary at these things. But to be AGI, it has to have some level of competency when used on problems outside its training set. In particular, it the LLMs were to install a known chess engine and run that to get the moves when asked to play chess, that would qualify for more AGI-like behavior. But really, chess is such a simplistic game that they should be able to do decently well at it even without even needing that. At the very least, they should be able to consistently play without making illegal moves - something that many 7-year olds manage quite well.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#283

Post by phist_mcgee »

yshklarov wrote: post #209

Great article, but the lack of sentence capitalization makes it unnecessarily difficult to read. Apologies if this comment is off-topic, but it really is quite egregious, and since the article was submitted by the author I presume they are open to the feedback.

I checked their whole site, it's all capitalised this way.

It's definitely... unique?

Re: Why I'm still bearish on LLMs after Navier-Stokes

#284

Post by user43928 »

21asdffdsa12 wrote: post #205
joefourier wrote: post #31

Earlier quoted context omitted.

> current frontier models > Gemini 2.5 Pro, O3, Claude Sonnet 3.7 and ChatGPT 4.1 The gap in capabilities between those models which they tested, and actual current frontier ones is enormous. I would not trust that any conclusions they made are applicable.

So give me a falsifiable point in time, a model you would claim succeeds at the task. One does not get to hotfix-patch updater out of the pressures of reality. Today is the day.

There is no need to ask. If you want to test SOTA models today, there are obviously only two: GPT-6 Astra and Fable 5.1.

The models listed in the paper are from early 2025 and are no longer relevant, much less on the frontier.

That Claude version is no longer available today, Gemini 2.5 Pro will be shutdown next month, and the OpenAI models are only available via the API today.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#285

Post by sfn42 »

Onavo wrote: post #202
minraws wrote: post #64

Earlier quoted context omitted.

I know HN readers and posters just read numbers and can't be bothered to read, but please read the methodology before making any claims. > About their ELO ratings from their own website: > A field-relative rating calculated within ChessBench. It compares performance among the tested models and is not a direct equivalent of a human chess rating. I am around 1600 elo in over the board I can mop up Astra Fable etc even…

> even if I give them literal infinite time and all the subagents and internet access.. Don't use the word infinite in any CS claims. They can recreate or approximate monte Carlo tree search and it technically is still a correct solution in your framing of the problem so long they defeat you.

[flagged]

Re: Why I'm still bearish on LLMs after Navier-Stokes

#286

Post by foldr »

thom wrote: post #276
foldr wrote: post #267

Earlier quoted context omitted.

Humans don't need a lot of training and finite tuning to make only legal moves. An intelligent adult could simply read a short summary of the rules of chess and then, if they were careful, play a very bad game of chess without making illegal moves. An LLM that has not been trained on any chess data cannot do that, at present. If you doubt it, take a current model and tell it that you want to play it at a variant of c…

How long a prompt do you think would be required to cajole an LLM into making legal moves at the rate of a human? Or do you think no amount of prompting could do that?

I don't know. My understanding is that current models will eventually fall into making illegal moves in longer chess games, and that no amount of prompting reliably gets them to stop doing so.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#287

Post by TheOtherHobbes »

linkjuice4all wrote: post #158
xdavidliu wrote: post #128

Earlier quoted context omitted.

that is if it even a human commenter at all

State-sponsored psyop meta comments aside, the models obviously continue to get better, but there is still a lot of 'guard railing' required to keep even the latest models completely on-task. The chess example is interesting because it's clearly a well-studied and established domain so the rules, strategies, and whatever else is in the training data should make yield excellent results; but clearly there is some behav…

I'm not sure why anyone is expecting stochastic systems to be deterministic.

Chess is a deterministic game won by a combination of known movesets and constrained multi-level forward search.

LLMs do neither of these things. They don't reproduce training data exactly, their next response is more 'inspired by' prompts and its own memory than produced deterministically, and they don't have the capability to do general forward search on their own.

So when you ask an LLM to play chess you're getting the equivalent of a very compressed and lossy JPEG of chess rules and strategies with added per-turn random noise.

They also don't have the ability to design their own chess engine, although it would be interesting to see what happens if you ask for one.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#288

Post by lelanthran »

TrackerFF wrote: post #196

The challenge with estimating abilities, is that we don’t know what the models can achieve if we just burn enough money. The navier-stokes shows us what mathematical problem can be solved when $10m worth of compute is thrown at something. It also makes one wonder: What could AI solve if we managed to orchestrate billions worth of agents to take on a specific problem? IMO the very best case scenario / potential for th…

> The navier-stokes shows us what mathematical problem can be solved when $10m worth of compute is thrown at something.

> It also makes one wonder: What could AI solve if we managed to orchestrate billions worth of agents to take on a specific problem?

We need to have robotics automation catchup first. The math and coding problems are problems in written-space only: you can set up feedback loops to test what worked and what didn't, then try to resolve the defects, maybe back up and try a different path, etc.

What solved coding and maths problems weren't the damn models; open up a chat interface to a SOTA model and you'll see they are pretty limited in producing a solution without a feedback loop.

Instead, it was the harness around the models: it let them explore a space and use feedback to control and direct that exploration.

Until we can do it in meatspace, it's kinda pointless sinking a ton of money into large problems facing mankind...

Like establishing a colony on mars (so the next rock to hit earth isn't an ELE).

Or moving us to a post-scarcity utopia, ending the concept of money.

Or designing and building better batteries for transport that uses only electricity (so that we stop using fossils as fuel).

Or actually building mass-housing. Or mass-farming. Or both, potentially ending homelessness and starvation.

Those are all worthwhile problems to solve, but where's the point of getting a solution on paper? There's no exploratory mechanism there, even for humans, to come up with a solution.

So, all we are left with then is making knowledge workers obsolete: another ELE, but of a different, self-inflicted kind.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#289

Post by zahlman »

RealityVoid wrote: post #247
dgb23 wrote: post #237

Earlier quoted context omitted.

The gap in capabilities is mostly quantitative and not qualitative.

Is it? I am on the fence on this, but it does seem like there are some qualitative improvements between the models. Not related to your post, but a fact I keep mulling over. The fact I don't trust the current crop of LLM's enough and I consider LLM's as a tech will hit a ceiling pretty hard, it doesn't mean parallel improvement curves won't spring up out of other research that will lead to much higher capabilities th…

> but it does seem like there are some qualitative improvements between the models.

It could easily seem that way, I think, in a "quantity has a quality of its own" kind of way. When you can come to the same conclusion faster, that lets you iterate more; and sometimes when you iterate you find more things.

Re: Why I'm still bearish on LLMs after Navier-Stokes

#290

Post by lelanthran »

intrasight wrote: post #258
dotdi wrote: post #211

Earlier quoted context omitted.

The whole point of this post was that it's questionable what can be achieved without huge investments into oversight and steering, because navier-stokes was a topic with an unusual level of specification. The problem itself was a specification. Such situations are rare in real-world scenarios. AI agents are good at solving well-specified tasks, not at solving problems. They do well in fields where the cost/effort of…

> a topic with an unusual level of specification Solving cancer also has an unusual level of specification. Many real world problems have that characteristic.

> Solving cancer also has an unusual level of specification.

Where did you read that?

"Cancer" is not just a single disease, even though we layman use the term that way. Cancer is a family of diseases, each probably having their own specific solution, but even in each of these individual diseases, there is no specification at the level of any open maths problem.

Post reply on HN