Research · Updated 14 Oct, 11:45 pm IST
Study: Frontier AI models rapidly improving at theoretical science problems
Why it matters for readers: This indicates AI is moving from pattern-matching to solving new, sophisticated scientific questions.
- Researchers reported that the best AI models went from solving ~2% to ~35% of theoretical problems between mid-2024 and mid-2026.1
- Fields highlighted as likely to be affected include probability, statistics theory, economic theory, high-energy theory, gravitation and cosmology, and quantum physics.1
- The study evaluated multiple OpenAI frontier models (o1, o3, GPT-5, GPT-5.1, GPT-5.4, GPT-5.5) on research problems posted online between December 2025 and May 2026.1
- Solutions found by the models corresponded to papers published after the training cutoff for GPT-5.5, indicating capabilities beyond memorized content.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started