Where AI should and should not be used in a science course
Short answer: in a science course, keep AI out of the practical itself (the observation is the evidence), the individually assessed write-up, and closed-book tests. Use it where it sharpens thinking: critiquing an investigation design the student drafted first, surfacing misconceptions, comparing explanations, and auditing an AI-written explanation against what students actually saw.
Why science is a clear case
Science courses are built on first-hand evidence. If students let an AI tell them what an osmosis investigation would show, they skip the observation the rest of the unit depends on. At the same time, science has rich places for AI: designs to critique, explanations to check, data to organise.
One course, four units
This is Biology 10, the real course Compound Intellect is tested against. The scheme of work has four units:
| Unit | What the scheme of work says | Kept AI-free |
|---|---|---|
| Cell Structure and Transport | Students design and carry out an osmosis investigation and write it up individually. | The investigation (curriculum) and the write-up (school policy) |
| Energy in Living Systems | Enzyme rate practical with an assessed write-up. | The assessed write-up |
| Inheritance and Variation | Ends with an independent closed-book test. | The closed-book test (school policy) |
| Ecosystem Stability and Change | Students collect local population data and present an evidence briefing. | Decided per activity by the teacher |
Around those boundaries, the same course uses AI for generating starter questions, comparing explanations, surfacing misconceptions, preparing worked examples, varying practice prompts and organising evidence.
Where AI helps in science
- Critique a design (Teach with AI): students write their own controls first, then ask only for variables they may have missed.
- Audit an explanation (Teach about AI): students check an AI-written mechanism against their own observations and mark where it is wrong.
- Surface misconceptions: AI-generated wrong answers make good discussion prompts.
- Teacher preparation (Teacher-use only): differentiated scaffolds, worked examples and practice questions.
Where it does not
- The practical observation itself.
- Individually assessed write-ups, where the school's policy prohibits AI in summative work.
- Closed-book tests of recall.
Related: The four ways to use AI in a classroom · Protecting assessment from AI
Questions schools ask
- Should students use AI for science practicals?
- Not for the practical itself: the observation is the evidence the unit builds on. AI can help before it, critiquing a design the student drafted, and after it, as something to check against what they saw.
- Can AI be used for science lab reports?
- If the report is individually assessed and the school's policy prohibits AI in summative work, no. In practice tasks, AI feedback on a draft the student wrote first can help.
- What is a good AI activity in a science lesson?
- Ask students to audit an AI-written explanation of a process against their own observations. The AI's mistakes become the lesson.
The pilot is one department, one term, free and with no card. You keep every plan, package and resource whatever you decide afterwards.
Talk about a pilotbasharat@compound-intellect.com