OpenAI Codex independently derives Jacobian conjecture counterexample
It replicated a recent human discovery without web search.
Combined views
138.3K
12 Sources, first seen ago
Sources
SB 
Sebastien Bubeck@SebastienBubeckyes GPT can do it too ;-). I highly recommend looking at Codex's write-up, its reasoning to find the counterexample is superb. So much for "the illusion of thinking" 😅. Also @aaron_lou's tweet includes the exact prompt he used on our internal Codex.…
- likes: 470
- replies: 26
- bookmarks: 122
- reposts: 28
JT Jerry Tworek@MillionIntInteresting what’s special about that specific counterexample https://twitter.com/aaron_lou/status/2079218392452530249
- likes: 139
- replies: 5
- bookmarks: 34
- reposts: 3
BH ben hylak@benhylakit's kinda crazy that, given how much openai spends, they don't regularly try disproving open conjectures? https://twitter.com/aaron_lou/status/2079218392452530249
- likes: 52
- replies: 4
- bookmarks: 4
- reposts: 0
RA rohan anil@_arohan_This one has a prompt too! https://twitter.com/aaron_lou/status/2079218392452530249
- likes: 57
- replies: 3
- bookmarks: 3
- reposts: 0
🇺🇦A 🇺🇦 Alex Polozov@SkiminokThe best instant advancement to human mathematics would be sharing the full reasoning trace that produced this counterexample. Alas, we won't get it because that model is closed and its creator company hates open-source AI. Don't worry. Open models will repro it soon enough.…
- likes: 77
- replies: 17
- bookmarks: 5
- reposts: 1
CP Chris Paxton@chris_j_paxtonJagged AGI at work. At some point it stops becoming news that another theorem was proven/disproven. They are better at math than we are. https://twitter.com/__alpoge__/status/2079028340955197566
- likes: 31
- replies: 1
- bookmarks: 2
- reposts: 1