First Proof
2026/02/05 by Mohammed Abouzaid, Andrew J. Blumberg, Martin Hairer +8 · 23 voices
#cs.AI #math.AG #math.CO #math.GT #math.HO #math.RA
paper · pdf
Abstract
To assess the ability of current AI systems to correctly answer research-level mathematics questions, we share a set of ten math questions which have arisen naturally in the research process of the authors. The questions had not been shared publicly until now; the answers are known to the authors of the questions but will remain encrypted for a short time.
Discussions
- First Proof [hn, 186 points, 122 comments]
- Ten math problems with proofs known the authors. Proofs are encrypted until Feb 13. For all problems, authors claim both AI-based literature searches and zero-shot attempts at proofs failed. If you wa [bsky, 27 points, 0 comments]
- AIs can "pass the BAR exam" or write amazing code, you say? Duh, that's because that data exists and they have been trained on it. MIT, Stanford etc tested newest models on unpublished, research-level [bsky, 12 points, 1 comments]
- Die mathematischen Fähigkeiten von LLMs wurden anhand von zehn Problemen aus laufenden Forschungsprojekten getestet. Diese Aufgaben waren zum Zeitpunkt des Tests nicht öffentlich zugänglich und gehört [bsky, 11 points, 1 comments]
- First Proof [lobsters, 5 points, 0 comments]
- To assess the ability of current AI systems to correctly answer research-level mathematics questions, we share a set of ten math questions which have arisen naturally in the research process of the au [bsky, 4 points, 1 comments]
- El reto fue organizado por un grupo de matemáticos profesionales, que seleccionaron una serie de problemas surgidos de su propia investigación, resueltos por ellos pero no publicados hasta ahora, con [bsky, 2 points, 1 comments]
- First Proof [hn, 2 points, 1 comments]
- I’m still long on popcorn futures. arxiv.org/abs/2602.05192 [bsky, 1 points, 0 comments]
- First Proof https://arxiv.org/abs/2602.05192 https://news.ycombinator.com/item?id=46924591 [bsky, 0 points, 0 comments]
- First Proof https://arxiv.org/abs/2602.05192 [bsky, 0 points, 0 comments]
- "First Proof" was calibrated to be on the boundary of what commercial LLMs can do in one shot without intervention. DeepMind got 6/10 with a specialist team under unverified conditions and best-of-two [bsky, 0 points, 0 comments]
- The paper "First Proof" explores current AI capabilities in solving complex math by presenting ten previously undisclosed questions. The authors know the answers, but they remain encrypted temporarily [bsky, 0 points, 0 comments]
- First Proof https://lobste.rs/s/vqjoe3 #pdf #math #vibecoding [bsky, 0 points, 0 comments]
- can LLMs really solve novel math problems? “This manuscript represents our preliminary efforts to come up with an objective and realistic methodology for assessing the capabilities of AI systems to au [bsky, 0 points, 0 comments]
- ⚡ Hackernews Top story: First Proof [bsky, 0 points, 0 comments]
- https://arxiv.org/abs/2602.05192 「First Proof」というタイトルの論文。 AIシステムが研究レベルの数学問題を正しく解答できる能力を評価するため、研究過程で作成した10問の数学問題を提示。 これらは未公開で、解答は問題作成者には既知のものであるが、一定期間は暗号化された状態で保持されます。 [bsky, 0 points, 0 comments]
- First Proof https:// arxiv.org/abs/2602.05192 # arxiv [mastodon, 0 points, 0 comments]
- First Proof https://arxiv.org/abs/2602.05192 (https://news.ycombinator.com/item?id=46924591) [bsky, 0 points, 0 comments]
- First Proof https://arxiv.org/abs/2602.05192 (https://news.ycombinator.com/item?id=46924591) [bsky, 0 points, 0 comments]
- https://bsky.app/profile/hackernews.com.web.brid.gy/post/3mec2iikdhhh2 [bsky, 0 points, 0 comments]
- First Proof [bsky, 0 points, 0 comments]
- First Proof #HackerNews https://arxiv.org/abs/2602.05192 [bsky, 0 points, 0 comments]
Related