Almost everyone has asked a chatbot a question by now. ChatGPT, Gemini and Copilot answer in seconds, in smooth and confident language. Most of the time, that answer helps.<\/p>\n\n\n\n
Sometimes it is completely wrong, and the tone never changes. This is called AI hallucination. In 2023, a New York lawyer used ChatGPT<\/a> to find court cases for a legal filing. The chatbot gave him case names and quotes. He even asked whether the cases were real, and ChatGPT said yes. The cases were fake, and the judge sanctioned him and his law firm.<\/p>\n\n\n\n A year later, Google\u2019s AI Overview told a user to add non-toxic glue to pizza sauce<\/a> so the cheese would stick. It had copied that tip from an old Reddit joke.<\/p>\n\n\n\n These examples show why AI-generated answers need verification. For students using chatbots for assignments, coding or career research, knowing how hallucinations occur is the first step toward using AI responsibly.<\/p>\n\n\n\n