How AI gives teachers time back
Teaching is bottlenecked by preparation and feedback, not by knowledge. A typical week disappears into planning lessons, building differentiated materials, and marking. AI can take the first-draft burden off all three, often saving several hours a week — time you redirect into the parts only a human can do: knowing your students, reading the room, and responding to the individual in front of you. The goal is not to automate teaching; it is to automate the paperwork around it so you teach more and type less.
Lesson planning and differentiation
The strongest use is generating and adapting materials. Give the model the year group, learning objective, time available, and any constraints, and ask for a structured lesson plan with a starter, main activity, and plenary. The real power is differentiation: ask for the same worksheet at three reading levels, or with extension questions for the most able and scaffolding for those who need it. Always treat the output as a draft — read it for accuracy, adjust the tone for your class, and replace any examples that do not fit your context. Clear, specific prompts produce far better results, so it is worth learning the basics of prompting (see How to Become a Prompt Engineer).
Marking and feedback
AI can speed up feedback dramatically when used responsibly. Provide a rubric and an anonymised sample answer, and ask the model to draft feedback against each criterion — then edit it before it reaches the student. Use it to produce a bank of common-misconception comments you can reuse, or to rephrase harsh feedback more constructively. The non-negotiable rule: do not paste identifiable student data into consumer tools that may train on it. Anonymise work, or use a platform your institution has approved under your local data-protection rules.
Designing AI-resistant assessment
If students have AI, your assessments must assume they will use it. Rather than banning it, redesign tasks so generic AI output cannot earn the marks. Effective approaches include in-class writing and problem-solving, oral defences of submitted work, drafts and reflections that show the process, and questions tied to specific class discussions or local context the model cannot know. You can even use AI productively here — generate a draft answer yourself to see what a student might submit, then design questions that go beyond it.
Doing it safely and well
Three principles keep AI use professional. First, you are accountable for everything that reaches a student — always review, never copy blind, because models can state confident falsehoods. Second, protect data by anonymising and following policy. Third, be transparent with students and colleagues about how and where you use AI; modelling honest, critical use is itself a lesson. Start small with one task — next week’s worksheets, say — measure the time you actually save, and expand from there.