1 - Start ~4 mins
You have already practised spotting when media is trying to persuade you. Today the source is a different kind: an AI chat tool, a large language model that writes answers that sound human.
Quick board prompt before we go further:
- One way AI chat tools help schoolwork
- One way they can mislead you in schoolwork
By the end of this lesson you will be able to say, in plain words, how these tools work, and you will audit a set of invented AI answers using three calls: trust, verify, or reject.
2 - Examine and Discuss ~8 mins
A large language model (LLM) is the engine inside most AI chat tools. In plain terms, it has read huge amounts of text and learnt patterns in how words tend to follow each other. When you ask a question, it predicts the most likely next stretch of text, one piece at a time, until it has built a full answer.
That is why the answers often sound smooth, polite and confident. The model is good at sounding like helpful writing. It is not checking a facts database the way a librarian would, and it does not know when it is wrong in the way a person does. It can sound just as sure when it is right as when it has invented a detail.
Before you audit any answers, think for a moment: what would make you trust a smooth AI answer straight away, and what would make you double-check it?
3 - Hands-on: AI Answer Audit ~14 mins
Work through the interactive activity together. For each invented AI answer, weigh the tells (the clues in how the answer is written), then commit to one verdict:
- Trust it for now
- Verify before you use it
- Reject it: do not rely on it
After each call, read the reasoning and notice which tell did the most work for you.
AI answer audit
4 - Your Turn ~8 mins
For each AI answer you did not simply trust, write one concrete check you would run before using it. Before you finish, check:
- You have listed every answer you marked verify or reject (by number or short label).
- Beside each one, you have written one specific check (for example: look up the date in a second source, search for the named study, ask a teacher, open a trusted reference page).
- For at least one answer, you have named where you would look (a person, a type of site, or a kind of record), not only that you would check.
- You can say in one sentence why a confident tone is not enough on its own.
5 - Make Sense ~3 mins
Pull the threads together. A large language model predicts likely text from patterns. That is why it can sound calm and certain when it is helpful, when it is wrong, and when it has invented a detail.
Your job is not to ban the tool or to trust it blindly. Your job is to audit: trust when the answer is careful and checkable, verify when something specific needs proof, and reject when the tells pile up. The skill you practised today is the same critical habit you use on any other digital source, applied to a new kind of writer.
Unlock the full learning experience
You're previewing this lesson. Get full access to this lesson and hundreds more — each one ready to teach, with interactive activities, printable resources and pupil progress tracking built in.
This lesson is copyright of Coding Ireland 2017 - 2025. Unauthorised use, copying or distribution is not allowed.