Skip to content

Glossary / Artificial intelligence

Hallucination

Also called: AI hallucination

A hallucination is a false or invented answer that an AI presents with confidence, as if it were true.

An LLM does not know that it does not know. It produces the most plausible next words, and a plausible sentence can be false: a date, a quotation, a legal reference, a figure.

You reduce the risk by giving the AI the reference documents, which is the principle of RAG, and by asking it to cite its sources. You cannot remove it. Every important fact needs checking.

An example

Asked about a tax rule, an AI cites a legal article by its exact number. The article does not exist.

AR Digital’s advice

Ask the AI to cite its sources, then open every link. A link that leads nowhere is the sign of an invented answer.

AR Digital / Contact

A project behind this word?

Thirty minutes, by video call or at the agency, to go from the definition to your own case. We tell you frankly what we would do, at what price and with what funding.

Radu Ulea, founder of AR Digital

Radu Ulea

Founder, AR Digital

“We tell you what works, what doesn’t, and what it costs. Then we do it.”

hello@ardigital.eu