Study says AI proofs start from foundational axioms, not theorems
A preprint posted to arXiv in November 2026 compares how large language models write mathematical proofs with human proofs.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
A November 2026 arXiv preprint study finds that AI models build proofs from base axioms rather than established theorems. Mathematician Terence Tao has also discussed AI and the foundations of mathematics in an interview.
Why it matters
The study says this approach shifts how mathematicians evaluate proofs.
Coverage
- Math 2.0 | Los Angeles Review of Books (lareviewofbooks.org)
- Preprint Compares LLM Mathematical Proving Style With Human Proofs (casrai.org)
More from Technology & AI
- China puts humanoid robots to work serving food and sorting parcels
- Newsom proposes sweeping AI safety rules for Washington
- Google releases Gemini 4 Neon AI model with million-token context
- Microsoft and Intel widen their cloud chip partnership
- Google DeepMind rolls out Genie 4 to select subscribers
- AI helped AMD shorten chip design work, OpenAI says