AI tools for students
Eight weeks on using AI tools carefully: what they are good at, how they fail, and where using one is cheating.
- Grades
- 6–8
- Ages
- 11–14
- Length
- 8weeks
- Sessions a week
- 1
- Each session
- 75min
- Class size
- 12max
What this course is
Your child is already using these tools, or their classmates are. This course is an argument for using them deliberately rather than furtively, and it spends as much time on the failure modes as on the capabilities.
Students learn what a language model is doing when it answers — predicting plausible text, not looking something up — and then go and prove it to themselves by catching it inventing a fact, a book, and a citation. That single exercise changes how a thirteen-year-old treats these tools more than any warning from an adult does.
We are direct about academic honesty. The course states plainly where a tool is a legitimate aid and where using it is submitting work that is not yours, and asks students to write down their own line. We also cover privacy: what not to type into a chat box, and why.
Week by week
Every week ends with something that did not exist at the start of it. That is what is listed under “ends with”.
- Week 01
What these tools actually do
Prediction, training data and why a model sounds confident regardless of whether it is right. No mathematics, but no magic either.
Ends with: A written explanation, in the student's own words, of why a model can be fluent and wrong at the same time.
- Week 02
Asking well
Context, examples, constraints and iteration. Comparing a vague request against a specific one on the same task.
Ends with: A before-and-after pair showing a weak prompt and a strong one, with notes on what changed.
- Week 03
Catching it being wrong
Hallucination hunting. Students deliberately push a tool into inventing facts, dates and sources, and record what they find.
Ends with: A log of five confident errors the student found, with the correct answers.
- Week 04
Checking a claim
Verification habits: finding a primary source, spotting a fabricated citation, and knowing when to stop trusting an answer.
Ends with: A short fact-checked report on a topic of the student's choice.
- Week 05
Where the line is
Honest use versus dishonest use in schoolwork, discussed as a real question rather than a lecture. Students write their own rules and defend them to the group.
Ends with: A personal one-page policy on AI use in schoolwork.
- Week 06
Privacy and what not to type
What happens to text put into a chat box, personal information, and family or school data that should never go in.
Ends with: A checklist of what the student will and will not share with an AI tool.
- Week 07
Using AI on a real project
Applying it properly to something the student is actually making: planning, drafting, and getting unstuck on code, with the student staying the author.
Ends with: A project the student built with AI assistance, plus a written note of exactly where it helped.
- Week 08
Presenting and defending
Each student presents their project and answers questions on which decisions were theirs.
Ends with: A presentation the student can answer questions on.
What they will be able to do
- Explain in plain language what a language model is doing
- Recognise a confident but fabricated answer and check it
- Use AI as an aid without handing over the authorship of their work
- Know what personal information not to put into a chat box
What they build
- A documented log of AI failures the student found themselves
- A fact-checked short report
- A personal AI-use policy for schoolwork
- A small project built with disclosed AI assistance