What AI actually is, in one page

AI is a program that predicts what should come next. That single idea explains most of what these tools are good at and most of what they get wrong.

Most of the tools arriving in classrooms are language models. They were built by reading a very large amount of written text and learning which words and ideas tend to follow which others. They are not looking facts up in a database. They are producing the most likely next sentence, which is often right and occasionally confidently wrong.

What that means in practice

A tool that predicts text is genuinely useful for explaining one idea three different ways, turning a rough paragraph into an outline, or quizzing a student on spelling. The same tool will invent a source, misdate an event, or state something false in a calm, expert voice, because it is finishing a pattern rather than checking a fact.

Four things worth telling your student

1. It is a very well-read guesser, not an answer machine.
2. It can be wrong and still sound sure, so ask it to show its work.
3. It does not know your student, and it does not have your student's interests at heart.
4. Anything typed into it can be read by someone else, so treat the chat like a note you would hand to a teacher.

Why districts are writing rules about this

A tool's answer is only half of the question. The other half is what the tool collects, who can read the chats, and whether the district has an agreement with the company behind it. That is what a permission slip covers, and it is the part most families never see unless someone writes it down in plain English.

The short version to read with a younger student: this is a powerful helper with no judgement of its own, so an adult stays in charge of the decision.