Why AI Gets Things Wrong

Because the model is predicting likely text rather than checking facts, it can produce something that sounds right and is completely made up. People call this hallucination. It shows up most with specific details: names, numbers, dates, quotes, sources, and anything obscure. The model has no internal alarm that says 'I do not know this.' It just keeps producing likely-sounding text.

A second limit is the context window, which is how much text the model can hold in view at once. Long conversations or huge documents can push early details out of reach, and the model may quietly ignore or blur them.

A third is that the model tends to go along with the framing you give it. Ask a leading question and you often get a leading answer.

Takeaway: use AI to draft, explain, and brainstorm freely, and verify anything that carries a name, a number, a date, or a real-world consequence.