Researchers Identify Flaws in AI Safety Benchmarks, Urge Reform
Experts have uncovered significant weaknesses in the benchmarks used to evaluate the safety and effectiveness of artificial intelligence (AI) models. A study conducted by a team of computer scientists from…