Iro AIWhat's in the kit
- A five-lesson sequence you can run in a week or spread over a term
- Five activities that need no software or student logins
- A free 10-question pre and post check that takes about two minutes
- A four-line process log for any assignment where students use AI
- A simple permission scale so students know when AI is allowed
Before the unit: a two-minute baseline
Have students take the free AI IQ test. It is 10 questions, takes about two minutes and needs no account. Note each student's score, then have them take it again at the end of the unit.
For context, 196 people who took the test online between April 1 and October 8, 2026 scored a median of 7 out of 10, and about 1 in 7 got a perfect score. That sample was self-selected and is not representative of students, so treat it as a rough reference point, not a benchmark. Full data: the average AI literacy score.
The five lessons
- How these systems work. Language models predict likely text rather than look facts up. That is why they can invent a convincing citation.
- Prompting as specification. A good prompt gives context, a role, a goal and a format. Treat it as a writing skill.
- Verification. Models sound just as sure when they are wrong. Students learn to check claims against primary sources. This is the most valuable lesson in the unit.
- Limits and failure modes. Where AI is reliably weak (recent events, some calculations, finding sources, high-stakes errors) and where it is strong.
- Ethics and disclosure. Bias in training data, what happens to what you paste in, and how to disclose AI use honestly.
Five activities, no accounts needed
1. Hallucination hunt
Bring in AI output with errors. Students find each one and prove it wrong.
2. Prompt A/B
Students compare the output of a vague prompt with a well-specified one and explain the difference.
3. Source-check relay
Teams trace four factual claims in an AI-written paragraph to primary sources and label each confirmed, unconfirmed or wrong.
4. Improve the draft
Students rewrite a flat AI-written piece for a specific reader.
5. Where AI is the wrong tool
Students sort real tasks into "AI does this well", "AI helps but needs checking" and "don't use AI here", then defend their choices.
Assessing it
Assess what students do, not what they remember. Can they write a specific prompt, catch a planted error, judge whether AI fits a task, and say honestly how they used it?
The four-line process log. Ask for it with any assignment where AI is allowed:
- What I asked
- What it gave me
- What I changed
- Why
A note on AI detectors. Detectors produce false positives and flag non-native English writers more often. Use a detector score, if at all, only to start a conversation, never as proof. More in how to tell if something is AI-generated.
A permission scale for assignments
Make AI use visible and graded instead of banned. Put one of these on each assignment:
- No AI. Work is entirely the student's.
- AI to brainstorm. Ideas only; the writing is the student's. Log required.
- AI to draft, student to verify. Any AI text must be checked and changed. Log required.
- AI encouraged. Use it freely, disclose how. Log required.
After the unit
Have students retake the test at tryiro.com/quiz and compare scores with their baseline. Students who want to keep going can use Iro on their own. All 34 learning paths are free, with one new lesson a day.
What Iro is and isn't for classrooms
Iro is a self-paced app for individual learners aged 13 and up. It has no teacher dashboard, rostering, assignment flow or LMS integration, and there is no school licence. This kit works without it. The full picture is on our page for educators.