A preliminary version of my book "The distribution of prime numbers" can be now downloaded for free, with the permission of the AMS: https://t.co/xZxMlYTTi8. If you enjoy the book, consider buying it from them: https://t.co/1bx2Zxe51r.
6/6 AI is transforming science. We have a responsibility to ensure that its fundamental research infrastructure is developed as a public good and governed with the scientific community in an ethical and accountable manner.
1/6 AI is here to stay. We cannot ignore these tools; we have to fully incorporate them into scientific research. We urgently need publicly funded, freely accessible GPT/Codex-type tools backed by frontier models and substantial public compute.
5/6 This is not only about making AI cheaper for academics. Scientific research should not become dependent on the priorities, products, and business models of a handful of private companies.
Peter Koymans, Mark Shusterman and I have just posted our resolution of the probabilistic version of Chowla’s non-vanishing conjecture for imaginary quadratic characters over Fq(T), with q ≡ 1 mod 8:
https://t.co/UuFAVCEHpb
(Over function fields, one cannot expect deterministic non-vanishing: Wanlin Li constructed infinitely many examples of vanishing)
We have been working on this for seven years, and it is very exciting to finally be able to share it!
@yiannissak I don't think we'll be writing proofs the same way anymore. Even if humans do all the high level job, AI will still be writing substantial pieces of the proof. To trust its writing, Lean certificates will be crucial.
Like many others, I had envisioned Lean would free us from the tedious work of verifying technical details, letting us focus on big ideas. Instead, its main role will likely be to certify AI-generated math, so we can safely invest human effort in understanding and rewriting it.
@EMehmetKiral Yes, I don't think humans will be writing Lean code. We just need to know enough to be able to check the statement the Lean proof is supposedly verifying.
There is a super shady data control setting on everyone's ChatGPT which makes it seem like your data can be used for training even if you explicitly say to *NOT* improve the model for everyone (1/9)
There is a super shady data control setting on everyone's ChatGPT which makes it seem like your data can be used for training even if you explicitly say to *NOT* improve the model for everyone (1/9)
Mathematicians aren't going to lose their jobs because of AI. Nor their prestige. Nor the fun of finding things out.
The only thing they'll lose is their chance to be heroes.
https://t.co/9NPR4Zdbwh
I’m deeply saddened to hear that Florian Luca has passed away. I will remember him as a kind and generous person, full of ideas and energy. Rest in peace.
Using mostly my voice, I solved one of @EpochAIResearch's FrontierMath Open Problems: finding an explicit presentation of the 2-adic Absolute Galois Group - open for more than forty years, now with a full proof in collaboration with David Roe, the problem's proposer. 🧵 1/n
Ten selected design proposals for the next series of euro banknotes have been unveiled.
Explore the proposals and tell us which ones you prefer.
Have your say https://t.co/VvcczFFxX3