Jun 12
We are broadening the scope of the blog posts on MathArena, might want to extend it further to allow for all kinds of analyses on AI4Math, not sure how exactly it should look like though.
1
6
634
Jun 12
Pythagoras-Prover: Advancing efficient formal proving via augmented Lean formalisation. ~ Joshua Ong Jun Leang, Zheng Zhao, Mihaela Cătălina Stoian, Qiyuan Xu, Haonan Li, Wenda Li, Shay B. Cohen, Eleonora Giunchiglia. arxiv.org/abs/2606.12594v1 #AI4Math #LeanProver #ITP
1
8
554
Jun 12
Auto formalisation of Chaitin and of the surprise incompleteness theorem. ~ Thierry Coquand. arxiv.org/abs/2606.12462v1 #Agda #ITP #AI4Math
5
9
812
Jose Brox 🏳️🌈🏳️⚧️ retweeted
Jun 11
AI scores a ‘C–’ on its hardest math test yet. ~ Joseph Howlett. scientificamerican.com/artic… #AI4Math
1
4
486
Petr Misan retweeted
Jun 10
A golden age of maths is dawning and mathematicians are freaking out. ~ Alex Wilkins. newscientist.com/article/252… #AI4Math
3
8
855
Jun 10
Artificial intelligence for mathematical reasoning: an integrated survey of language models, neuro-symbolic systems, and verified discovery. ~ Syed Rifat Raiyan, Mohsinul Kabir, Hasan Mahmud, Md Kamrul Hasan. arxiv.org/abs/2606.08728v1 #AI4Math
2
16
559
Jose Brox 🏳️🌈🏳️⚧️ retweeted
Jun 10
How Terry Tao became an evangelist for AI in math. ~ Kevin Hartnett. quantamagazine.org/how-terry… #AI4Math #LeanProver #ITP
1
7
504
Jun 10
Les mathématiciens se positionnent face à l’IA. ~ François Lassagne. pourlascience.fr/sd/mathemat… #AI4Math
1
3
618
Jun 8
The next way to build agents is to have them design their own graph structure. We demonstrated it for AI4Math. That is just one use case.
Hitting SOTA with all open-weights was truly amazing—and it shows that OSS models are great.
Amazing to work with such brilliant people!
Jun 8
Introducing Goedel-Architect: an open-source framework for formal theorem proving in Lean 4. Using the open-weight DeepSeek-V4-Flash (284B-A13B), it reaches state-of-the-art results, rivaling proprietary systems at a fraction of the cost.
It solves 4/6 on IMO 2025, 11/12 on Putnam 2025, and 3/6 on USAMO 2026. On PutnamBench it solves 88.8% (597/672) at just ~$1.65 per problem.
Paper: arxiv.org/abs/2606.06468
Project page: goedelarchitect.github.io/
4
51
Jun 7
It is not easy to open-source things at Google, at least not quickly. I am not aware of any Google entry, but I don't know much about the other teams working on AI4Math (though they would all encounter the same logistical difficulty).
1
134
1
3
156
A problem of Andrews and Dhar on partitions. ~ Simon Mahns, Ken Ono, Jujian Zhang. arxiv.org/abs/2606.05117v1 #AI4Math #AxiomProver #LeanProver #ITP
9
586
DeepMind’s new AI found a strange new way to think. ~ Károly Zsolnai-Fehér. youtu.be/Dkqzqw8rxXI #AI4Math #LeanProver #ITP
7
746
Very interesting talk by @JaumedeDios at the @CRMatematica organized by @matdepuab.
"The goal of a PhD is to learn, papers are by-products" (my transcription)
cc @therfer #AI4math
1
2
10
488