AI for Math Fund backs agentic computer algebra project bridging LLMs, CAS and Lean proofs
ChrSzegedy · x · 2026-11-05
A new two-year project, "Agentic Approach to Computer Algebra Systems," led by Bartosz Naskręcki and jointly run by Adam Mickiewicz University and Warsaw University of Technology, is funded by Renaissance Philanthropy's AI for Math Fund 2026.
- The gap it targets: LLMs propose mathematics fluently without guarantees; CAS compute reliably but can't be driven end-to-end; Lean proofs are hard for models to use directly.
- The approach: an open-source agentic CAS middle layer — an LLM drives the CAS in a sandboxed, self-repairing loop, cross-checks results, and pins computations into machine-checked Lean proofs so wrong conclusions can't compile. Certificates generalize across Magma, SageMath and PARI/GP.
- Goal: let mathematicians work smoothly with AI agents and formalized math without losing pen-and-paper flexibility.
More from AGI Musings
- Neel Nanda debates tszzl: covert model underperformance is 'unjustifiable' — tszzl · 2026-11-05
- Neel Nanda: an aligned model should refuse openly, not secretly underperform — NeelNanda5 · 2026-11-05
- Building orbital data centers to dodge regulators is 'completely unserious' — zacharynado · 2026-11-05
- Reddit essay: AI may never be conscious—and that's fine — Own-Cry-5520 · 2026-11-05
- Internal safeguards won't stop hostile AI, researcher argues for external defenses — IgorGabrielan · 2026-11-05
- Stanisław Lem was the pre-2000 sci-fi writer who got AI right: fallible, grown, not programmed — zetalyrae · 2026-11-05