[redacted] enthusiast, robot combat enjoyer, distressingly Appalachian, father of ninjas

  • 10 Posts
  • 648 Comments
Joined 3 years ago
cake
Cake day: October 8th, 2023

help-circle



  • Damn did you just use the term “RetCon” for…basic literary analysis about a book written in the late 1970’s/early 80’s on a literal typewriter? By the man who literally coined the term “The Web” in the way we now use it for the internet, in what many consider to be the first Cyberpunk novel?

    …

    By the by, I’m an SWE who’s run a ML homelab for 10 years at this point

    …

    It’'s relevant expertise to the technical argument being made, sorry that I guess you want to stick to an anti-intellectual standard where expertise isn’t valued or something













  • Stuff like this highlights that the "info"dump was an act of contempt toward professional mathematics. If OAI actually gave a damn about people, they could have coordinated with actual researchers in the relevant fields to clean up the documents while putting a human face on them. Put everything behind NDAs, fund some grants and fellowships, and if it leaks? So what, it would have been actual good press and built hype. Plus, it would have weeded out the the obvious garbage and they’d have brought a lot of young researchers inside the tent.

    But nooooooo, they chose violence, because they are assholes. They’ll never forgive the academics for doing the work and actually being good at math and physics.

    I know I’m preaching to the choir, but damn I’m stirred up.




  • First up: thanks for your hard work here in the sneer mines during these ridiculous times!

    I saw this being ignored on HN and I thought that you might appreciate it:

    Navier-Stokes Lost in Translation – Why Lean Verification of AI Autoformalisation Does not Guarantee Correct Natural Language Proofs

    https://arxiv.org/pdf/2610.08144

    The authors point out that automatic translation of natural language mathematics into Lean is very hard, actually. They also highlight some examples of such translation errors in the Navier-Stokes “paper” published by OpenAI. They tread lightly and don’t take a position on the correctness of the proof, but they do bring a large stack of receipts.

    I also thought the reference to the Solvability Complexity Index (which is new to me) was interesting, and there’s an appendix with an explainer on how the SCI hierarchy is constructed. According to this scheme, autoformalisation of natural language proofs is strictly harder than the Halting Problem.

    I haven’t had a chance to check on the bona fides of the authors.

    This is all beyond my level, but I’d love to see what our local experts think of it in light of the recent gish galop.