GPT-6-Astra solves 49 of 50 physics research problems at $2.40 apiece
AI Summary
GPT-6-Astra achieved a 98% success rate on the TheoPhysEval benchmark, solving 49 of 50 unpublished theoretical physics problems with accuracy matching leading specialists at $2.40 per problem. While the AI demonstrated exceptional problem-solving capability and even caught errors in expert reference solutions, researchers note it still lacks the "taste" needed for breakthrough discoveries that would parallel recent advances in AI-assisted mathematics.
From the source
On the TheoPhysEval benchmark, GPT-6-Astra solves 49 of 50 unpublished theoretical physics calculations as accurately as leading specialists. In some cases where the agent seemed to fail, the experts' own reference solutions turned out to contain errors. But the researchers say physics hasn't had a breakthrough like the ones in mathematics, because the agents lack "taste." They still think autonomous, top-level AI research could arrive soon. They already see consequences for how young physicists
The full text couldn't be loaded here (the source may require a subscription).
View original at The Decoder