AIHacker Newsabout 5 hours ago
Expert Re-Grading Finds Physics Benchmarks Flawed
3 min read
A new study by over 50 authors found leading physics benchmarks for AI models are broken. Experts re-graded responses and found most 'errors' were actually flawed questions or reference solutions, not model failures.
Level
Hype check
Tap to vote and see what everyone thinks.