Why AI Confidently Makes Things Up

The trickiest thing about AI is that it can be completely wrong while sounding completely sure. There is no nervous tone, no maybe. It states a made-up fact as smoothly as a true one.

These confident errors are called hallucinations. Understanding why they happen is your first defense against getting fooled.

Remember how AI works: it predicts likely words, one after another. It is not looking up facts in a database. It is generating text that sounds right.

Most of the time, sounding right and being right line up. But sometimes the most natural-sounding sentence is simply false, and the AI has no built-in alarm telling it so.