Your goal: assemble a subject lesson and an interdisciplinary extension, then evaluate the pupil learning rather than the polish of the AI-assisted output.

Differentiate support and challenge

Keep the learning intention visible while varying text complexity, representations, scaffolds or constraints. A pupil may explain with a diagram, speech or an assistive communication system. Additional challenge can mean defending a choice, finding a counterexample or considering another perspective—not simply doing more questions. Avoid using AI to assign fixed ability labels or infer needs from identifiable records.

Gather evidence before, during and after

First elicit an initial attempt so you know what pupils already bring. During the activity, observe a decision, question, explanation or revision. Afterward, use a fresh independent task, with ordinary accessibility supports, to see whether the intended reasoning transfers. The term independent means without AI supplying the target thinking; it does not mean removing necessary accommodations.

A pupil may submit an excellent poster but be unable to explain its main claim. That is a reason to investigate what was learned, not automatically accuse the pupil of misconduct. Ask them to discuss a choice, check a source or revise a related example. Do not rely on an AI detector or the model’s impression as proof of authorship.

Responsible use as lesson design

Use fictional or non-identifying material in these exercises. Before actual deployment, check what the service collects, retains and shares, along with school rules and applicable permissions. Removing a name may not anonymise distinctive details. Review factual claims and how people or communities are represented. Provide an equivalent route without a paid tool or direct voice interaction. Explain what AI contributed and what pupils created or checked.

Use transparent language such as “The teacher used AI to propose two fictional arguments. We checked them and wrote our own conclusions”—only when that describes what happened. Pupils should know that a simulated character is generated, can be wrong and does not know their feelings. Keep responsibility for feedback, wellbeing and consequential decisions with accountable adults.

Final planning project

Combine one subject lesson with one interdisciplinary extension. Include the age band, intention, initial pupil thinking, limited AI contribution, verification, voice-supported option, offline alternative, inclusive participation, transfer task and observation criteria. Use the downloadable templates and your module notes. The written project is self-reviewed; the final scenario assessment tests decisions separately.

Review the value after teaching

Ask whether the AI contribution opened a useful question, made a comparison possible or supported revision. Compare the evidence with the learning intention and a reasonable non-AI alternative. If the activity mostly increased tool time or polished the product, redesign or omit that contribution. The course’s examples are planning proposals, not claims that AI automatically improves attainment.