@AndilesAnthony
Joined March 2022
Replying to @mikurosblues
🤖 Made with AI
14
3
1
33
17,399
I am on OpenAI’s side. They rightfully deserve credit for solving Navier Stokes. In particular , the strategy that was employed and bootstrapped generally follows the “procrastination “ strategy mentioned here: mathoverflow.net/questions/4… Congrats.
2
1
2
201
Look at Terry’s thought process. It basically foreshadows the solution methods employed by Levent and others. So OpenAI is not at fault.
1
63
Anyone wanting some real insights from Terry Tao on AI see this: teorth.github.io/tao-web/ai-… ACI is here but not AGI. @OpenAI @AnthropicAI @gdb @OfficialLoganK @littmath
1
78
Replying to @mikurosblues
🤖 Made with AI
14
3
1
33
17,399
Replying to @GaryMarcus
I am at the cusp of solving a problem whose solution, if written up, could be punished in the Annals of Mathematics (considered best math journal in the world). That sounds good but imo it’s not a great achievement. So far , I have had codex run for 2 weeks on 5.5 extra high on fast mode and it hasn’t been able to resolve the problem. It has access to ChatGPT 5.5 Pro and yet it can’t resolve it. This is a case where I believe either me giving a prompt that subtly suggests some vague idea or asking Fable 5 will definitely be beneficial. Fable 5 is good ChatGPT 5.5 pro is just coming about with the same conclusion written in a different way.
2
1
148
If anyone wants to see the nontrivial insight, please like this post.
1
81