What “AI hallucination” means

A simple explanation of AI hallucination and why it matters.

AI hallucination is when an AI system gives you something that sounds right, but isn’t. It might invent facts, misquote sources, or confidently return an answer that doesn’t actually exist. The problem isn’t just that it’s wrong, it’s that nothing in the response tells you that.

For example, an AI tool might summarise an article and include details that were never there, or provide a reference you can’t find anywhere. The output still reads well. It feels complete. That’s what makes it easy to miss.

This happens because AI models are designed to predict likely responses, not check whether something is true. They’re optimising for what sounds plausible, not what’s been verified.

Most of the time, that distinction isn’t obvious. If something looks reliable, we tend to treat it as reliable. That’s where it starts to matter.

So the question isn’t just whether AI can be wrong. It’s how often we notice when it is.