Studying With AI vs Letting It Study For You
The Fluency Trap
Reading a brilliant explanation feels like learning. It is not. That feeling is fluency — the ease of processing something clear — and your brain reliably mistakes it for competence. This is why you can nod along to a worked example and then freeze at a nearly identical exam question. AI produces exceptionally fluent explanations on demand, which makes the trap deeper than it has ever been. The test is simple and slightly uncomfortable: close the tab and reproduce it. Explain it aloud, or write it out from memory. If you cannot, you did not learn it, however good the explanation felt.
- Understanding an explanation and being able to produce it are different abilities
- Fluency feels like mastery, which is exactly why it fools careful people too
- The check is retrieval: close everything and reproduce the idea unaided
- Struggling for a bit before you get help is not wasted time — it is the part that builds the memory
What AI Is Genuinely Excellent For
Used as a study tool rather than an answer machine, it is close to unfair how useful it is. It will generate unlimited practice questions at the exact difficulty you ask for. It will mark your attempt and explain where your reasoning broke rather than just what the answer was. It will explain the same concept five different ways until one lands. It will play the examiner, or the sceptic who attacks your essay argument. It is available at eleven at night when nobody else is. All of these keep you doing the cognitive work while removing the bottleneck of waiting for someone to help.
- Endless practice questions and past-paper-style problems, tuned to your level
- Feedback on your reasoning, which is more useful than the correct answer alone
- Re-explanation from a different angle when a textbook has lost you
- Adversarial practice: ask it to attack your argument and find your weakest point
The Line: Who Is Doing the Thinking?
Forget rules for a second and use one question. If you ask AI to generate practice questions, you do the thinking. If you ask it to check your reasoning, you did the thinking. If you ask it to write your conclusion, it did the thinking. If you ask it to explain a comment your teacher made, you are doing the thinking with better information. The tool is not the variable — the same app is a tutor or a substitute depending on where the cognitive effort lands. Notice which one you are choosing, especially at eleven at night when the deadline is tomorrow and the honest answer is uncomfortable.
- Ask before each use: after this, will I be able to do it myself?
- Tutor pattern: AI asks you questions, checks your work, explains your mistakes
- Substitute pattern: AI produces the artefact and you tidy the wording
- Deadline pressure is exactly when the line blurs, so decide your rule in advance
Try It Yourself
The fluency trap is invisible from the inside, so the only way to find out whether you actually know something is to try to produce it. This takes about eight minutes and uses a topic you are revising anyway.
Pick something you are genuinely revising this week. Close everything and write down everything you can remember about it for two minutes — messy notes are fine. Then paste your own version into an AI tool and ask it to mark it, using the prompt below. You do the recalling; it does the marking.
I am revising [topic] at [your year or level]. Below is everything I could remember without notes. Mark it: what I got right, what I left out, and what I got subtly wrong. Do not rewrite it for me — just tell me where the gaps are, and ask me one question about the weakest bit. [paste your two minutes of recall here]
- Your two minutes of recall was thinner than you expected before you started
- It named at least one thing you had left out completely
- You can now say which specific part to go back over, instead of "revise the whole topic"
Prefer slides, quizzes, and saved progress? Read this lesson in the library — free, no sign-up.