Five questions to ask before bringing an AI tool into your school or college

Most leaders I speak to now have a steady stream of AI products arriving in their inbox, each one promising to save time or raise standards. Some of them will do exactly that. Others will create more work than they save, or bring risks that nobody spots until it's too late. Telling the two apart isn't easy when every demo looks polished, so these are the questions I'd want answered before signing anything. They apply to any AI tool, including ours.

1. What problem does it actually solve?

It sounds obvious, but it's the question that gets skipped most often. A tool should be fixing something your staff or learners genuinely struggle with, not something that looked exciting in a demo. If the supplier can't tell you how you'd know, six months in, whether it's working, that's a gap you'll end up having to fill yourself.

2. What happens to your data?

Ask exactly what the tool collects, where it's stored, how long it's kept, and whether any of it is used to train the AI model behind it. The DfE's guidance is clear that personal data has to be protected, and that responsibility doesn't pass to the supplier just because you've signed a contract. If the answers are vague, or need a lawyer to decode, treat that as an answer in itself.

3. Does it need learner information?

This deserves a question of its own, because data about children and young people carries a much higher bar. Some tools genuinely need learner information to work and some don't, and you should know which kind you're buying. Where it isn't needed, it shouldn't be collected. We built EducAitor® so that it doesn't store, identify or use any learner information at all, because feedback on teaching simply doesn't require it.

4. Who checks what it produces?

AI gets things wrong, and sometimes it does so very confidently. The DfE's guidance says teachers should verify what AI produces and that learning should stay teacher-led, so it's worth asking exactly where a person sits in the process. A tool that produces something for a teacher to review is very different from one that sends its output straight to learners, parents or managers without anyone looking at it first.

5. Is it honest about the AI underneath?

Many education products are built on AI models from other companies, which is perfectly normal, but staff should be told. When we submitted our app, Apple required us to make it clear to users that we use third-party AI, and that's a reasonable expectation of any supplier. If a product is cagey about what's powering it, you're entitled to wonder what else it isn't telling you.

The answers matter more than the demo

Any supplier worth working with should be able to answer all five of these plainly. If they can't, or the answers seem to shift depending on who's asking, that's worth paying attention to. AI can do a great deal of good in education, but only when the people using it can see clearly what they're being asked to trust.

Sources

• DfE publishes guidance for use of AI in schools (UKAuthority, June 2025)