TOPICS / RESOURCES

AI literacy · 7 min read

How to Tell When AI Is Making Things Up

A practical checking routine for detecting unsupported, distorted, or overconfident AI answers.

Watch for unnecessary precision

Exact dates, percentages, quotations, and named studies can create an impression of authority. Precision is not proof. Treat a detail as a claim that needs verification, especially when it is central to your decision.

Separate explanation from evidence

A coherent explanation can still be wrong. Ask which statements are directly supported, which are interpretations, and which are guesses. Then verify the important claims using primary or reputable sources.

Check the source, not just the citation

A citation-shaped answer is not the same as a useful source. Open the original page, confirm that it says what the answer claims, and check whether the context changes the meaning.

Ask for uncertainty explicitly

Request a list of assumptions, disputed points, and information that would change the answer. A reliable workflow makes uncertainty visible instead of rewarding confident language.