Skip to main content
AI News List

List of AI News about AI in mathematics

Time Details
2026-01-22
13:46 UTC
Test-Time Training and Open Source AI Models Drive State-of-the-Art Discoveries in Science: Business Opportunities in Mathematics, Algorithms, and Biology

According to Stanford AI Lab (@StanfordAILab), leveraging test-time training combined with open-source AI models now enables researchers and businesses to achieve state-of-the-art (SOTA) scientific discoveries with only a modest investment. This approach surpasses prompt engineering methods used with closed frontier models like Gemini and GPT-5 for complex discovery tasks in mathematics, kernel engineering, algorithms, and biology (source: Stanford AI Lab, Twitter, Jan 22, 2026). The practical implication is a democratization of advanced AI capabilities, lowering costs for organizations aiming to innovate in scientific research. This trend opens up significant business opportunities for startups and enterprises to build specialized AI solutions for scientific domains, using open models and custom training to outperform proprietary alternatives.

Source
2026-01-17
20:03 UTC
GPT-5.2 Pro Solves Open Erdős Problem #281: Major Leap for AI in Advanced Mathematics

According to @gdb and @neelsomani on X, GPT-5.2 Pro has successfully solved Erdős problem #281, a previously unsolved challenge in mathematics, without any prior human solutions. Mathematician Terence Tao described this as 'perhaps the most unambiguous instance' of AI independently solving an open mathematical problem (source: https://x.com/neelsomani/status/2012695714187325745). This achievement demonstrates the rapidly growing practical potential of advanced AI models in theoretical mathematics and scientific research. The use of GPT-5.2 Pro in this context highlights new business opportunities in AI-driven mathematical discovery and automation, as well as future trends in AI-assisted scientific innovation.

Source
2025-12-10
16:01 UTC
Axiom Achieves Breakthrough Math Results Using ThinkyMachines Tinker for AI Research Infrastructure

According to @soumithchintala, Axiom, an AI research lab launched just four months ago, achieved remarkable results on the Putnam math competition by leveraging the Tinker infrastructure platform from ThinkyMachines (@thinkymachines). By using Tinker to rapidly bootstrap their AI research workflows, Axiom's autonomous AxiomProver system solved 9 out of 12 Putnam problems in Lean, a performance that would have ranked #1 among around 4,000 participants last year and placed them as a Putnam Fellow in recent years (source: @soumithchintala, Dec 11, 2025; @axiommathai). This serves as a concrete early validation that Tinker could become for AI frontier research labs what AWS was for product startups in the 2010s, potentially transforming how AI teams access scalable, specialized infrastructure to accelerate mathematical research and innovation.

Source