How to use AI to learn, not to avoid learning

AI is now part of every student’s toolkit, whether universities like it or not. The interesting divide is no longer between students who use it and students who don’t. It’s between the two ways it gets used — because the same tool can deepen your understanding or quietly replace it.

The trap: outsourcing the thinking

Asking AI to produce the thing — the essay, the summary, the answer — feels productive. Work appears. But the struggle it skips is where learning happens: retrieving what you know, noticing what you don’t, forcing vague ideas into precise sentences. Skip that repeatedly and you end up submitting work you don’t understand and revising notes you never made.

The exam hall is where the bill arrives. No tab to open, no prompt to type — just what actually settled.

The test: one question before every prompt

There's a simple way to tell which side of the line you're on.

Before any prompt, one question is enough: am I using this to understand something, or to avoid understanding it? The first kind of use leaves you knowing more after the chat than before. The second leaves you with an artefact — and nothing else.

The honest answer varies by moment, not by person. The same student who uses AI well on Tuesday can use it badly on Thursday night with a deadline looming. That's why it works better as a habit of checking than a rule you set once.

Using AI to support your learning

Used deliberately, AI can be one of the most powerful learning tools students have ever had — not because it produces work, but because it can question, test and correct you on demand, at any hour, for as long as you need. Four ways in which AI can actually support learning rather than replace it include:

1. Getting useful explanations

When a lecture explanation doesn’t land, ask for another — simpler, from a different angle, or built on something you already understand. You can keep asking follow-ups until it clicks, which is exactly what a textbook can’t offer and a lecturer rarely has time for.

2. Tracing the logic

Have AI walk a proof or argument one step at a time, and stop it wherever you can’t say why the step follows. That “why does that follow?” moment is precisely the gap a lecture glosses over — and precisely what an examiner tests.

3. Applying concepts to practice

Ask AI for worked examples: how a concept plays out in a real case, a real dataset, a real problem — then try the next one and have AI check your attempt.

This is where understanding gets tested. A definition you can recite is not the same as a doctrine you can apply to a set of facts you haven't seen before. Working through examples — and getting your attempts corrected as you go — closes the distance between knowing what something says and knowing how to use it.

4. Finding gaps in your understanding

Describe a topic from memory, then ask what you missed or got wrong. It’s revision and diagnosis in one: the material you couldn’t recall is exactly the material that needs another pass.

The tool isn’t the risk

AI won’t make understanding optional. Exams, interviews and the work after graduation still test the person, not the tools. Used to avoid learning, AI is an expensive way to stay exactly where you are. Used to learn, it’s the most patient tutor you’ll ever have.