Why AI makes things up, explained for parents and children
Children (and adults) often ask why an AI tool says something confidently and then turns out to be wrong. The short answer is that these tools are built to write likely text, not to look up true facts.
Guessing the next word
A language model learns patterns from a very large amount of text. When you ask it something, it builds its answer one small piece at a time, each time choosing a word that fits well after what came before. Try finishing this sentence: “The cat sat on the…”
Figure 1 · Choosing a likely word
This works well when the pattern is common. It works less well for rare facts, recent events, or exact details, such as a date, a number, or the title of a book. The tool still produces something that sounds right, because sounding right is what it is good at.
Three habits children can learn
- Ask “how would I know if this is wrong?” Before using an answer, think of one way to check it.
- Check with a second source. A textbook, a teacher, or a trusted website can confirm a fact.
- Ask for the steps. For maths or reasoning, ask the tool to show how it got there, then see if each step makes sense.
A good activity at home
With your child, ask an AI tool a question you already know the answer to, and see how it does. Then ask about something specific and obscure that you can look up, and compare. Talk about what you notice. The aim is not to make children afraid of the tools. It is to make them good at checking.
A note on privacy
Children should not type personal details, such as full names, addresses, school names, or photos, into an AI tool. Check the age rules of any tool before your child uses it.
Checking answers and using AI carefully are part of our student program.