What are AI 'hallucinations' and can we fix LLMs so that they don't happen?

This video explains why large language models generate false or nonsensical outputs called AI hallucinations, which appear plausible but are factually incorrect. Mark Russinovich discusses the fundamental technical reasons LLMs make these mistakes and what strategies developers can employ to reduce hallucination rates. It is aimed at developers working with generative AI who need to understand these limitations when building AI-powered applications.