AI has started hallucinating less often, but it still confidently passes off lies as truth

▤ Summary

This article examines how large language models still produce false statements even as hallucinations become less frequent. It references Axios and a Boston Consulting Group report on the ongoing accuracy gap in AI systems. It also mentions retrieval-augmented generation as one way to improve responses, but notes that errors remain a real problem. The piece centers on AI software capabilities and limitations, which makes it relevant to software industry readers.

▦ Linked Companies