MIT Study: AI Accuracy on Theory Problems Jumped from 2% to 35% in Two Years

An MIT CSAIL study spanning probability, statistics, economic theory, high-energy physics, cosmology and quantum physics found top AI models' success rate on frontier theory problems rose from about 2.8% to 34.6% between mid-2024 and mid-2026, suggesting AI is poised to disrupt theoretical science.

2026-10-15 ~ 2026-10-17 · 3 related posts

• MIT study: best AI models went from solving 2% to 35% of theory problems in two years — nordicinst · 2026-10-15
• MIT study: frontier AI solves cutting-edge theory problems at 34.6%, up from 2.8% in 16 months — MIT_CSAIL · 2026-10-17

1 near-duplicate retellings: mikeflache