Hybrid activities pair stages where students use AI on purpose with stages where the learning must come from them. The idea is easy to agree with. The hard part is knowing what it looks like in a real tutorial on a Tuesday morning.
Here are six patterns you can adapt this semester. Each one names its AI-integrated stage, its AI-resistant stage, and why the combination works. The examples come from health education, but every pattern transfers to other disciplines.
1. Find the flaw
AI-integrated: You give students an AI-generated answer that looks convincing but contains planted errors. For example, a medication plan with a subtle contraindication, or a discharge summary that misses a red flag.
AI-resistant: In class, students find, explain and correct the errors, ideally in small groups and without devices.
Why it works: It trains the most important AI skill there is: spotting confident nonsense. Students learn that fluent text isn’t the same as correct text, and they practise the verification habits their profession will expect.
2. Human first, AI second
AI-resistant: Students draft their own response first, in class or under light supervision. A rough handwritten plan is fine.
AI-integrated: They then ask an AI tool to critique their draft and suggest improvements.
What they submit: The final version plus a short change log: what they accepted from the AI, what they rejected, and why.
Why it works: Students’ own thinking comes first, so AI becomes a coach rather than a ghostwriter. The change log makes their judgement visible and gives you something genuinely worth marking.
3. Defend it
AI-integrated: Students may use AI freely to help produce a piece of work, such as a health promotion proposal or a literature summary.
AI-resistant: A short oral check follows, of around five minutes per student or group. You ask a few targeted questions: why this approach, what would change for a different population, and where the weakest evidence sits.
Why it works: A brief conversation reveals understanding faster than any detection tool. Students who used AI well can talk confidently about their work. Students who didn’t will find that hard, and they’ll know in advance that the conversation is coming.
4. Anchor it locally
AI-integrated: Students use AI to research general background, such as the evidence base for falls prevention.
AI-resistant: They then apply it to something AI cannot know: a de-identified situation from their placement, data collected in class, or an interview with a peer or practitioner.
Why it works: The analysis depends on information only the student has. AI can help with the general picture, but the application has to be theirs.
5. Adjudicate the experts
AI-integrated: Students prompt AI to argue two opposing positions, or you provide two AI responses that disagree. For example, conflicting recommendations on early mobilisation after surgery.
AI-resistant: Students judge between them using primary evidence they locate and appraise themselves, then defend their verdict in a tutorial discussion.
Why it works: It builds evaluative judgement and shows students that AI can sound authoritative on both sides of the same question.
6. Simulate, then debrief
AI-integrated: Students practise a difficult conversation with an AI playing a patient, family member or colleague. Breaking bad news, managing a complaint and taking a history all work well.
AI-resistant: A live, observed debrief or role-play follows, where they apply what they practised with a real person and get feedback.
Why it works: AI offers safe, unlimited rehearsal. The observed stage confirms the skill transfers to real human interaction, which is what professional accreditation cares about.
Making any of these work
A few principles apply whichever pattern you choose:
- Label every stage. Tell students plainly whether AI is expected, allowed or not allowed at each step, and why.
- Match the pattern to your delivery mode. Fully online units have fewer options for resistant stages. Patterns 2 and 3 adapt best, using timed drafting and video-call orals.
- Mark the process, not just the product. If your rubric only grades the final document, students will optimise for the document. Give real weight to the change log, the defence or the debrief.
- Be honest about resistance. Nothing is completely AI-proof. Explain to students why each checkpoint exists, and they’ll be far more likely to engage with it in good faith.
Start with one pattern in one unit, see how students respond, and build from there.
Our Hybrid Activity Designer builds draft activities like these for you: tell it your learning outcome, discipline and delivery mode, and it designs the stages, the student instructions and the marking guidance.