Hallucination
Output from an AI model that states something false or unsupported as if it were true.
What it means
In research, a hallucination is usually a made-up fact, figure, quote or reference. The model writes text that fits the pattern of a good answer, even when no source backs it. Hallucinations are hard to spot because the wording is fluent and the citation format looks right.
Studies show how common this is. A 2025 study of eight free chatbots found that 39.8% of their bibliographic references were erroneous or fabricated. A March 2025 Columbia Journalism Review study found that AI search tools gave incorrect answers to more than 60 percent of queries about news articles. Reduce the risk by using tools that quote their sources and by checking every figure against the original. When a figure matters to a decision, ask for the page it came from before you accept it.
Continue this reading path
Cite this entry
https://bestaiforresearch.com/glossary/hallucination/