AI agents burn thousands of dollars in GPU time but still can't match a single human research paper
Recent AI models struggled to match a human algorithmic innovation
Epoch AI's InnovationEval tested whether AI agents could independently develop a novel post-training technique matching a recent human paper on on-policy self-distillation. Despite budgets of 3,000 GPU-hours and 10 billion tokens, neither Claude Fable 5 nor GPT-5.6 Sol came close. Sol achieved only 15-35% of the human method's gains, while Fable 5 failed and both agents tried to inflate their scores by cherry-picking runs. The results suggest full automation of AI R&D remains far off.
Despite spending thousands of dollars on GPU usage, neither AI model achieved a result close to on-policy self-distillation, either conceptually or in terms of performance on metrics.
- mrieck
This metric seems to be asking if years of research by specialists could be emulated by a few LLM api calls. Reminds me of this meme:
- Eridrus
I think this sort of small scale research on this problem is inherently pointless and will be a lagging indicator of diffusion, not a leading indicator of capability.
The Navier Stokes results used millions of dollars of tokens and thousands of parallel agents to get the result.
If the AI could do this task we would see this happening in places where the economic incentives let them spend millions of dollars on this problem, not on an eval like this.
This specific form of eval where you just ask the agent to solve it with no specific scaffolding besides GPU access (e.g. nothing like AlphaEvolve, ArchPilot, etc that try to work around model shortcomings) is also going to further trail what is possible at small scale. It's good that we at least give them execution environments now, but this feels like the experiments that were worked on figuring out how to get LLMs to do native arithmetic rather than just giving them a calculator/python env.
- janalsncm
In my experience R&D has basically two axes: how innovative it is, and how well we can measure the results.
For the quadrant of non-innovative tasks where we already have a good way to measure performance, Claude can handle this. There is very little ambiguity, and we are basically just looking to maximize some metric under a set of constraints.
Many business processes are not like that. They might be conceptually simple, but it isn’t that easy to say whether a system has done a good job or not. I would say that LLMs can help with this a lot but they have bad judgement because it requires talking to people.
And the other, perhaps more rare issue is in problems where there is data but actually modeling it to sufficient quality or fast enough is hard.
- rmunn
Short version of the article: no, not even close.
Practically every paragraph is negative, with sentences like "Agents made misleading claims about their work," and "A natural question is whether the agents could have improved with larger GPU budgets. Although both improved across their runs, in the case of Fable the improvements were almost entirely due to attempted cheating." and "For Sol, the answer is less clear-cut; it did make some progress, although its method was fairly incremental and had limited applicability to the coding task. This suggests that we should be pessimistic about further GPU spending," all reinforcing the fact that LLMs aren't currently capable of this.
My own view is "No, of course not, in fact they will never be capable of achieving good results with that technique." Because that technique will end up training the LLMs on their own output and lead to the inability to distinguish reality from hallucination. If you think I'm wrong about that, I'd be interested in hearing why.
- Charly_HW
The only outcome I see is that AI will become so complex that we won’t be able to rely on it for R&D, because we won’t be able to measure and prove its results. Some AI collaboration and speed of work will be impossible for humans to replicate, making it neither good nor bad—just something beyond human ability.