Large models suffer from hallucination issues, which are particularly difficult to detect in learning and Q&A scenarios. Users point out that in non-coding fields like basic subjects or industry information, learning relies on the correctness of model outputs. However, models like Gemini often lack citation sources and search functionality, leading to low credibility. Through practical experience sharing, the author suggests repeatedly emphasizing source requirements, prompting the model to provide cited websites for users to verify independently. This discussion reveals practical coping strategies for AI hallucinations, helping users improve information accuracy, and holds significant reference value for AI learners and developers.
