Why does AI sometimes make things up?
Sample lesson from the Office management Learning curve, "Why AI sometimes talks nonsense with full confidence".
Hallucination: an answer the AI gives with confidence, but which is made up or incorrect.
An AI is trained like a pupil taking a multiple-choice test with no penalty for wrong answers. A guess sometimes scores a point, a blank never does. That is why an AI would rather fill a gap than say: I do not know. That means it can tell you something with complete confidence that is simply wrong: a made up phone number, the wrong opening hours, a supplier that does not exist. This is called a "hallucination". It is not lying, it is guessing. It just always sounds equally certain, including when it is off the mark.
This week, ask something factual (an address, a price, a date) and check the answer against the real source. That way you feel for yourself how convincing a mistake can sound.
This lesson appears in the newsletter at three levels, Starter, Practical and In-depth. Same explanation, more or less detail; switching is free.
This is a sample lesson
You get the whole Learning curve (with examples from your own field, at the level you choose) in inklaretaal's weekly newsletter. The first month is free.