When AI Gets It Wrong

AI makes mistakes — sometimes hilarious ones! Understanding why helps you use AI more wisely.

Why AI makes mistakes

AI learns from examples. If those examples have mistakes, the AI learns the mistakes too. And sometimes AI encounters situations that are quite different from anything in its training — so it guesses, and guesses wrong.

Famous AI blunders

Early Google Translate once converted "time flies like an arrow; fruit flies like a banana" into something completely nonsensical — because it could not work out whether "flies" was a verb or a noun.

Image recognition AI famously labelled some people incorrectly because of biases in the photos it was trained on — a serious problem that researchers are still working to fix.

AI chatbots have confidently cited books, studies, and famous people saying things that never happened.

Mistakes with consequences

Most AI mistakes are harmless and funny. But some mistakes matter. If AI is used to make important decisions about people — like whether someone gets a job or a loan — mistakes can be really unfair. That is why humans always need to stay in charge of important decisions.

The good news

Researchers are always working to make AI more accurate, more fair, and less likely to make these kinds of mistakes. And having people like you who understand AI limitations helps!