What Is an AI Hallucination?
An AI hallucination is an output from an artificial intelligence system, most often a large language model, that is false, fabricated or unsupported, but presented fluently and confidently as if it were true. The word is a metaphor, because the system is not perceiving things. It is generating text that fits patterns without checking accuracy.
Common examples
- Invented references: realistic-looking citations with authors, journals and dates that do not exist.
- False quotations attributed to real people.
- Wrong facts, such as dates, statistics or names.
- Made-up details about events, laws, products or research.
- Incorrect code or calculations that look reasonable.
- Misreading documents, adding claims not in the text.
- Confident answers to questions that cannot be answered.
Why it happens
Language models predict plausible text from training patterns. They do not have a built-in database of verified facts or a sense of truth. If a prompt asks for something the model does not know, or if the training data is incomplete or contradictory, it may produce a convincing but incorrect response. Longer answers, obscure topics and requests for exact citations raise the risk. Tools that search the web or documents reduce some errors but can still misquote or misinterpret sources.
Why it matters for students
Hallucinations can lead to false information in essays, invented references that count as fabrication, poor decisions and academic integrity problems. Using unchecked AI output in graded work can have serious consequences.
How to check output
- Treat claims as unverified until you check them.
- Find the original source for facts, numbers and quotations.
- Search for every reference in a library database and open it.
- Compare with course materials, which are the standard for your assessments.
- Ask the tool for its sources, then verify them.
- Repeat the question differently, and note inconsistencies.
- Use human experts, such as instructors and librarians, for important questions.
Reducing the risk
- Give clear context and ask it to say when it is unsure.
- Provide the source text and ask questions about it.
- Avoid asking for citations from memory.
- Use tools designed to cite real documents, while still verifying them.
Beyond text
Image and video generators can also produce errors, such as distorted details or impossible scenes, and can be used to create misleading media. Verify images used in work.
Related terms
See large language model, peer review, citation style, doi and primary source.
Frequently asked questions
Why do AI tools hallucinate?
They generate likely-sounding text from patterns, without a built-in way to check facts. When information is missing or unclear, they may fill gaps with plausible but false details.
How can I spot a hallucination?
Check facts, quotations and references against reliable sources. Be suspicious of very specific details with no source, references you cannot find and answers that change when you ask again.
Can hallucinations be eliminated?
They can be reduced with better training, retrieval from trusted sources and careful use, but not fully eliminated. Always verify important claims.