AI Models' Dirty Little Secret: Right Answers, Wrong Sources
A new benchmark reveals that even top-performing AI models often provide correct answers while citing incorrect sources, exposing a critical flaw in their ability to provide trustworthy information. This weakness has significant implications for industries that rely on verifiable data, such as finance and healthcare.
Leading AI models like GPT and Gemini routinely cite text passages in document analyses that don't actually support their answers. Even when the answer is right, the cited evidence is often wrong. Researchers at Peking University call this "attribution hallucination," a risk for regulated fields like law and medicine. Their new CiteVQA benchmark is the first to test for it systematically. The article AI models often give the right answers but point to the wrong sources appeared first on The Decoder.
