Musing · kept 5 November 2025
AI as a research partner
Timothy Gowers on using AI:
🧵 Full thread: https://x.com/wtgowers/status/1984340182351634571
I crossed an interesting threshold yesterday — one I suspect many mathematicians are now crossing. In the middle of trying to prove a result, I found a statement that looked true and would be useful if it were. Instead of proving it myself, I asked GPT-5, and in 20 seconds it produced a proof using a lemma I hadn’t heard of. I’m confident I’d have got there eventually, but it would’ve taken me about an hour. It feels like we’ve entered a brief but wonderful era where our research is greatly sped up by AI — but AI still needs us.
❦
☞ Threads from here
Nearby leaves, connected by subject and form.
094 The guy who financed Johannes Gutenberg another musing 092 AI blind spot another musing 089 Invention of modernity another musing 099 Did we get evolution wrong? another musing 088 Can AI eliminate information advantages another musing 100 AGI will be business as usual? another musing
browse all subjects →