Why Does AI Hallucinate? The Real Reason Chatbots Guess

A chatbot made up six court cases, and a lawyer filed them in federal court. Why do AI models state false things with total confidence? This video explains the two real reasons: how language models learn from text, and how we grade them on tests. You will also see the fix that researchers now propose. You will learn why "Adam Kalai's birthday" trips up chatbots, how a trick from Alan Turing predicts what a model cannot know, why leaderboards reward guessing over "I don't know", what Anthropic saw inside a model when it hallucinated, and how changing the grading can make models more honest.