Learning with AI and technology#
Ask an AI assistant to explain the French Revolution and you’ll have a clear, patient, tailored answer in seconds. Ask it to write your essay on the French Revolution and you’ll have that too — and you’ll have learned nothing. Both took the same ten seconds. That gap, between the tool that helps you think and the tool that thinks for you, is the whole story of learning with AI. It’s the most powerful study aid ever handed to ordinary learners, and the easiest one to use in a way that leaves you no better off. This section is about staying on the right side of that line.
The technology is genuinely new. The principle for using it well is not. Everything earlier in this manual — retrieve instead of reread, space your practice, wrestle with difficulty rather than smoothing it away — still decides whether you actually learn. AI doesn’t change those rules. It just makes it far easier to either honour them or sneak around them, depending on how you point it.
The opportunity: a tutor, a quizmaster, and a scheduler in your pocket#
Used well, AI removes three things that have always slowed learners down: waiting for an explanation, finding someone to test you, and organising your own review. That’s a lot of friction to delete.
-
Explanations on demand, at your level. Stuck on a step? Ask. Then ask it to explain the same thing again more simply, or with a different analogy, or as if you were ten. A good human tutor does this and most learners never get one. The thing that used to require a person and an appointment now requires a sentence.
-
Endless retrieval practice. This is the big one, because retrieval practice is the single highest-value move in the whole manual. Ask the AI to quiz you on a chapter, to generate practice problems, to keep asking until you can explain a concept back without peeking. You can manufacture as many tests as you have patience for — and testing yourself, not rereading, is what builds durable memory.
-
Feedback in the moment. Try to explain something in your own words and ask the AI where your explanation is thin or wrong. Attempt a problem first, then check. Catching a misunderstanding the same minute you have it, rather than weeks later on an exam, is exactly the kind of fast feedback skilled learning runs on.
-
Scheduling and planning. Hand it your goal and your deadline and let it draft a spaced review schedule , break a daunting subject into a sensible order, or simply nag you to come back tomorrow. This is offloading done right — you’re outsourcing the admin of learning, not the learning.
Notice the pattern in everything above. The AI is generating questions for you to answer, explanations for you to wrestle with, schedules that send you back to do the work. The effort stays where it belongs: in your head.
The central risk: cognitive offloading#
Here’s the trap, and it’s a subtle one because it doesn’t feel like a mistake while you’re making it.
Psychologists call it cognitive offloading — using the world, or a tool, to reduce the mental work you’d otherwise have to do (Risko & Gilbert, 2016). Most of it is harmless and sensible. Writing a number down instead of holding it in your head, setting an alarm instead of remembering, using a calculator for arithmetic you understand — that’s just good sense, and humans have always done it. The problem is narrow but serious: when you offload the exact mental effort that would have produced the learning, the tool does the work and you keep none of it.
That’s the difference between a calculator and an answer key. A calculator spares you arithmetic you already grasp so you can think about the harder thing. An answer machine that hands you the solved problem spares you the struggle that was the lesson. The struggle wasn’t in the way of the learning. The struggle was the learning — this is the desirable difficulty principle, and AI makes it dangerously easy to skip.
And the cost may not stop at “you didn’t learn this chapter.” Some early studies suggest heavy reliance on AI tools may go hand in hand with weaker critical-thinking habits, with the effect running partly through increased offloading — the more people let the tool think, the less they practise thinking themselves (Gerlich, 2025). Treat that as a caution rather than a verdict: it’s recent, correlational, and a young field. But it points exactly where common sense already does. A skill you never exercise doesn’t hold.
The tell is how an interaction feels. If the AI did something you couldn’t yet do yourself and you simply accepted the output, you offloaded. If it pushed you to do something you couldn’t quite do, and you did it, you learned. Same tool, opposite result.
The principle: more retrieval and feedback, not fewer hard parts#
So here’s the one rule to carry through everything that follows. Use AI to create more of the effortful work that builds learning — and never to skip it.
A simple test settles almost every case. Before you accept what the AI just gave you, ask: did this add a rep, or remove one?
- Make it ask, not tell. “Quiz me on this.” “Give me three problems like this one.” “Don’t show the answer yet — tell me if mine is right.” You want the AI generating the difficulty, not dissolving it.
- Attempt first, check second. Produce your own answer, explanation, or solution before you let the AI weigh in. The first pass from your own memory is the rep that counts; the AI is the marker, not the author.
- Use it to explain, then test — never to replace recall. Have it clarify what you don’t get, then close the chat and retrieve it cold. If you can’t reproduce it without the assistant, you haven’t learned it; you’ve borrowed it.
- Offload the admin, not the thinking. Scheduling, sequencing, formatting, reminders — hand all of that over freely. The understanding and the remembering stay with you.
Get this right and AI becomes the most patient study partner you’ve ever had: a tutor that never tires of your questions and a quizmaster that never runs out of them. Get it wrong and it becomes a very expensive way to feel productive while learning nothing. The tool is the same. The choice is yours, made fresh in every interaction.
Takeaway#
AI can give you a personal tutor, an infinite supply of practice questions, instant feedback, and a review schedule — gifts no previous generation of learners had. The danger is cognitive offloading: letting the tool do the very thinking that would have built the learning. The fix is a single question you ask before accepting any AI output — did this add a rep or remove one? Make it quiz you, attempt things yourself before you check, and offload the admin while keeping the thinking. Point AI at more retrieval and feedback, never at skipping the hard parts, and it becomes the best learning aid you’ve ever had.