you are viewing a single comment's thread
view the rest of the comments
[–] 11 points 2 days ago* (4 children)

I'm probably wrong in my thinking but if the LLM actually solved the problem I don't see the Issue with this? Other than credit being attributed to the model itself rather than the people who programmed it and the mathematicians who wrote the training data it ingested to resolve it

  • source
  • hideshow 4 child comments
  • [–] 32 points 2 days ago

    Because math also has internal politics. If an AI "solution" is presented then you lose justification to go after it. Worse, you may be expected to spend months unraveling slop math. So it's better to just not talk about it because it only hurts you.

  • source
  • parent
  • [–] 26 points 2 days ago (1 child)

    The issues with the AI math "solutions" seems to be that they get a correct answer in really annoying ways.

    One of the points of solving a problem for mathematicians is to find a clean simple path, or at least the cleanest and simple path they can given their approach.

    I definitely believe that this is very good since many times the pathway to the solution uncovers a ton of more interesting questions maybe finds a general solution to a problem set.

    All the AI ones so far have just been millions of lines of LEAN vomit that technically compile as a proof, but don't really have any clean approach.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 26 points 2 days ago

    It's the "show your work" thing of mathematics causing problems. AI has oodles of proofs instead of just the relevant ones, even if it is correct, it doesn't properly articulate how it is correct, which means the results aren't easily replicated, especially since if you asked an AI to give you the answer a second time, it would give different results.

  • source
  • parent
  • [–] 11 points 1 day ago* (last edited 1 day ago)

    what other people have said, but also in the practically of academia (ie, if you want to keep having a job) your research has to be novel, which is why phd students are told never to share ideas about work they haven't done yet; if someone else does it before you you doing it is no longer worthy of getting the phd or, if you're already employed, keeping your position. it's not a system that helps science but scientists are workers and generally like to eat

  • source
  • parent