For the past day, everyone's been talking about a letter signed by a long list of Fields Medal-winning mathematicians and published by Terence Tao (a mathematician at UCLA and a Fields Medalist himself, from 2006). The title alone is grim: *A Severe Misalignment of AI in Mathematics*. The authors admit LLMs have made real progress in recent months and can now solve major unsolved problems across different branches of math — like OpenAI's recent claimed solution to the Navier–Stokes problem. They're not calling for a ban on AI or denying its usefulness. But — in their view, solving problems is just a tool and a proxy for the real goal: conceptual understanding and generating new insights. Forget that, and the tool starts destroying the very goal it was meant to serve. Mass-producing true/false statements, the authors argue, might not enrich the mathematical landscape — it could deplete it. Process matters more than result.
It's interesting how closely this echoes (just in a calmer register) an idea creators of every stripe have been pushing for a while now: the result matters less than the process of getting there, and since AI can't replicate the process (and doesn't need to, to get the result) — AI should be banned from getting results.
Programmers, essentially, complain about the same thing — they value process too. Makes you wonder who's next: managers, for instance, are no less attached to their procedures, meetings, and tickets :)
The letter itself — terrytao.wordpress.com.
And, sorry, I can't not bring up my own old piece on process-people vs. result-people — forbes.ru.