Why do AI chatbots ‘hallucinate’? And why they will never be 100% accurate?

Artificial intelligence (AI) chatbots are increasingly being used for almost everything, from customer service to writing assistance and developers using them to generate codes. Yet, there is one puzzling flaw that persists — they present completely wrong information with absolute confidence. In the study titled: ‘Why Language Models Hallucinate’, authored by researches Adam Tauman Kalai, Ofir Nachum, Edwin Zhang all from OpenAI and Santosh S Vempala of Georgia Tech, released on September 4, 2025, together studied this issue in depth and published…