The term "AI hallucination" refers to a phenomenon where the output of an AI system significantly deviates from the trained data. This term gained attention due to its association with AI, despite "hallucination" traditionally being a state experienced by humans. Unlike humans, AI lacks emotions and consciousness. AI hallucination occurs when the generated output of an AI system is misleading but presented in a coherent and logical manner, resembling factual information. The occurrence of AI hallucination in generative AI systems depends on the quality of the training data used to build them. High-quality data is essential for generating reliable information. Algorithms in generative AI produce outputs based on probabilities and statistics rather than true understanding of the content. The phenomenon may also be influenced by the model and structure of the questions posed by users, as AI reactions can vary with different phrasing. A breakthrough technology called Prediction-powered inference (PPI) has been introduced to address AI hallucination issues. PPI helps correct model outputs and rectify potential errors. Experts like Michael Jordan, a prominent professor at the University of California Berkeley, believe that these models can answer various questions effectively.