Why AI makes things up, and how to check its answers
Sometimes an AI tool gives an answer that sounds convincing but is simply wrong. People call this a hallucination.
- A hallucination is a confident but false answer
- Invented sources and links are common
- Check anything important independently
- Ask the tool to flag uncertainty
Why it happens
A language model generates the most plausible-sounding text based on patterns it learned. It is not checking a database of facts, so when it lacks information it may still produce a fluent, confident answer. Invented quotes, statistics, book titles and web links are typical examples.
A simple checking routine
- Ask for sources, then open them yourself
- Cross-check key facts with official or primary sources
- Verify every number, date and quotation
- Be extra careful with legal, medical and financial topics
- Ask the tool to say when it is unsure
Reduce the risk
Give the tool the source text and ask it to answer only from that text. Keep questions specific, and use tools that can search and cite when you need current information.
The rule of thumb
The more it matters, the more you should verify. For anything you will publish, sign or rely on, a human check is essential.
Answer the question below. If you are not sure, say so. Then list the claims I should verify independently and what kind of source would confirm each one.
Frequently asked questions
Will AI stop hallucinating?
It may improve, but errors are unlikely to disappear entirely. Keep verifying anything important.
Do search-enabled tools avoid hallucinations?
They reduce the risk by citing sources, but they can still misread or misreport what they find, so open the sources.
AI tools, features, prices and policies change quickly, so check each provider鈥檚 current terms. Do not enter confidential or personal data into an AI tool unless you understand how it is used, and verify important facts independently.
