Teaching Students

More than eight in ten students use AI for schoolwork. About the same proportion say no teacher has ever shown them how.

Those two numbers describe a skill in universal use, taught by nobody, learned from peers and from the tool's own confident manner. Whatever your view about whether students should be using it, they are — which means the current arrangement is teaching them to use it badly.

For a workplace-oriented comparison outside education, this discussion offers another way to look at measurement, workload, behaviour, or accountability.

This section covers three separate things that are often merged and should not be.

The lesson. One session, any subject, built around a live demonstration of a confident wrong answer in your own subject. Failure lands where warnings do not, and asking the class how they would have caught it produces the checking strategies from them rather than from you.

For an external perspective on student digital literacy, privacy, and learning, see UNICEF.

The skill. Verification is the habit with the longest shelf life, because prompting technique is already changing and tool rankings will be different next year. The gap between a plausible answer and a correct one is permanent, and closing it transfers to sources, statistics and anything else that arrives looking finished.

The habit. Saying what you used is more durable than any rule about what is banned, and it rests entirely on one promise: honest disclosure of an over-the-line use is never punished. Break that once and no student discloses again.

Alongside those: what changes with age, when a student genuinely should not use it, and what should never leave the classroom.

A warning about the wider subject. Much of the comparison writing available online is produced by companies selling one of the tools being compared, and disclosure is inconsistent. Where we point at something, we say what it is and who makes it — and the resources we recommend most warmly are the ones run by charities and universities rather than by anyone with a growth target.

• Age Bands: 8, 13 and 17

  What actually differs by age, what the guidance does and doesn't cover, and why most age advice in this area is inference rather than evidence.

• The Skill of Checking the Output

  Verification is the one skill that stays useful whatever the tools do next. How to teach it, and how to make it assessable.

• Teaching the Habit of Disclosure

  Saying what you used is a more durable skill than any rule about what is banned. How to make it normal, safe and low-effort.

• Homework When the Answer Is Free

  Unsupervised work was always partly an honour system. What homework is still for, and what to set now that the answer is available.

• Student Data: What Leaves the Room

  Typing a student's name into the wrong tool is the commonest privacy mistake in schools. How the rules are structured and what to ask.

• Teaching a Class to Use It

  Almost every student uses it and almost none has been taught how. A lesson plan for showing them, including where it fails.

• When a Student Should Not Use It

  Not a list of banned tasks. A rule a student can apply themselves, and the four situations where the cost is highest.