What is 'hallucination' in the context of large language models like Claude?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Hallucination is like a student who doesn't know the answer but writes something convincing anyway — the output sounds authoritative but the facts are invented.
Full explanation below image
Full Explanation
Hallucination refers to a model generating information that is factually incorrect, fabricated, or unsupported by its training data, but presented with apparent confidence. This can manifest as fake citations, incorrect statistics, invented historical events, or plausible-sounding but wrong technical details. Option A describes refusals, which are an intentional safety behavior — the opposite of hallucination. Option C describes a vision processing error, which is a different type of error. Option D is an API infrastructure issue, completely unrelated to the model's output quality.