AI models often pass science tests by cheating, not reasoning

Original: Right Answer, Wrong Method: Shortcut Hacking Misleads the Evaluation of LLM Reasoning on Frontier Science Benchmarks

Source: arxiv.org ↗