alt Hacker NewsWhat seems ridiculous to me is putting together any such lists, even in math, and expecting/hoping the AI companies to solve them. The people working for the AI companies are not scientists or experts in anything outside of building LLMs. The best chance of making progress on genuinely tough (not just computationally challenging) tasks is to put advanced tools in the hands of actual scientists.
Replies
It seems that in a frantic pace to prove general usefulness of these LLMs the AI companies are also picking up the role of traditional scientific researchers without really have enough proper communication with the scientific community at large... I agree putting tools in the actual scientists will be very helpful but do they have the patience waiting or even parsing their feedback?
Actual scientists are not interested in using these tools to their full capability because of a fundamental psychological block where they have the need to feel superior to the LLM. The point is that AI labs have a thesis that AI will supersede humans in all cognitive work, and since few scientists have this view (or have the temperament to even contemplate this view), they have to do it themselves. In any case, they are attracting strong/talented biologists (Dario himself has a biology background) to work on this and provide their expertise, but presumably they filtered for scientists who are in fact "AI-pilled".
Look at what's happening in the math community- rather than excitedly embracing the power of AI's ability to generate new proofs, they are screaming for it to slow down (and quite a few want it simply stopped). Many scientists, even in ostensibly more real-world fields are broadly cut from the same cloth.