closing the loop on loop closing | founder & ceo @ContinuaAI | ex-Google Distinguished SWE | wheelie.dev
Brooklyn, NY
- Tweets 2,426
- Following 1,327
- Followers 4,159
- Likes 538
Wait a sec, didn't the committee advise "understanding" first?
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
159
i'd love to see the raw (unedited) thinking traces when i go ballistic and drop the f-bomb on my agents. i really think they're putting on a show. hell, show me the j-space.
117
If I put up a server to serve artisanal hand coded services for the discerning customer, does the server itself need to be certified organic code, too?
Fwiw, I feel about equally astonished at coding agents and the fact that we/humans built software pyramids stone by stone.
1
120
and that's the thing. look, i will always be a fan of Tao's artisanal Math, and his writing is so clear and brilliant. what a waste for him to be going through this crashout.
openai about to drop a math bombshell this week with around 400 AI-generated proofs potentially getting released all at once
terence tao is still trying to digest the Navier-Stokes example and he might be in for another crashout week
1
4
1,017
best thing ever: i'm working on drum part transcription software and my coding agent suggested we try "You Drink a Lot of Coffee for a Teenager". music.youtube.com/search?q=Y… do you think you could transcribe that?
1
226
death to banking and other websites that don't let you paste codes/account numbers/$$ into text fields.
1
210
With the global curiosity of Gemini 4, @OfficialLoganK , will one be able to use a subscription w/ 3rd party harnesses? (OpenAI allows that. Anthropic doesn't. What will you choose?)
207
Truth.
Reading all those recent beautiful essays by Terry Tao, Kevin Buzzard and @wtgowers, I could not resist, but write up my deeply personal perspective on how I perceive and see things:
docs.google.com/document/d/e…
3
464
every day i wake up and hope @earendilworks doesn't sell out. i have no evidence they ever were, but i hope they ride their /own/ rocket to the moon. bonus points for doing something good for the European tech scene.
167
i really am concerned. if this is the best the working group can come up with after weeks, and before that, months of "crisis" meetings, i think we need new mathematicians. (i live in absolutely respect and awe of that group and the broader community -- just see my bookshelf and youtube watch history -- which makes this all the more sad.)
Those recommendations are written as if humans will remain relevant in the production of mathematical results, and create an 'intelligibility tax."
This is unrealistic: AI math will be to human math what 1s of CPU arithmetic operations is to 1s of human arithmetic operations.
266
how we know Buckmaster's claims about OpenAI meddling are specious: @SchmidhuberAI hasn't taken on the case.
123
this is quite sad -- the biggest brains are on tilt. actually, it's quite a human story. the emotions of their "crisis" leading to irrationality.
I find this ridiculous: mathematicians "ask them to stop testing advanced mathematical problems on proprietary models."
I have bad news for them: whatever looks "advanced" math today will look utterly primitive to tomorrow's AI
1
249
