No AI Agent Passes More Than 46 of 168 Physics Research Attempts in New FrontierPhysics Benchmark
· techtimes.com · popularity 17/100 · 0 similar stories today · AI & Data Science (Artificial Intelligence)
Read the original at techtimes.com →
Summary
AI agents on physics research: FrontierPhysics, a new benchmark from UC Berkeley's Bingran You and 71 co-authors, found no agent passed more than 46 of 168 attempts. Claude Opus 5.5 led at 27.4 percent. The authors' failure review of GPT-6 Astra points to weak rigor and missing expert intuition, not gaps in physics knowledge.
Found in: Tech Times. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 14.