Explanation feels like learning; retrieval produces it. Ask to be tested before you ask to be told.
Do now: Attempt an answer first, then ask for feedback on your attempt rather than for the answer.
If you have not produced it from memory, you do not yet know it.AI is the most patient tutor most people will ever have, available at any hour, willing to explain the same thing six different ways. It is also the fastest route to the illusion of competence, because it removes the struggle that makes learning stick.
Attempt first, explanation second
The order matters more than anything else in this lesson. Attempting an answer badly, then being corrected, produces far more durable learning than reading a good explanation. The effort of retrieval is not an unfortunate cost of learning; it is the mechanism.
So invert the usual pattern. Instead of 'explain compound interest', try 'ask me three questions about compound interest, wait for my answers, then tell me exactly where my reasoning went wrong'. The instruction to wait matters, because a model will otherwise helpfully answer its own question.
When you are corrected, ask why your specific wrong answer was tempting. Understanding the shape of your own error is what stops you repeating it, and it is something a textbook cannot do for you.
Ask to be questioned, not told. And instruct it to wait for your answer.
Scaffolding you can actually use
Ask for an explanation at three levels: as if to someone who has never met the idea, as if to a colleague, and as if to a specialist. Where you first understand tells you your real starting point, which is often lower or higher than you assumed.
Ask for the misconception. 'What do people usually get wrong about this, and why is the wrong version appealing?' produces some of the most useful output available, because misconceptions are well represented in the material these systems learned from.
Ask for worked examples with the steps exposed, then attempt the next one alone before checking. Ask for a concrete case from your own context — a chama's savings, a shop's stock, a school's timetable — because abstract examples are harder to hold on to.
Space it out, and test yourself for real
Ask for questions about last week's material as well as today's. Revisiting after a gap, when it has become slightly difficult again, is what moves knowledge from short-term to durable. Sessions that only cover today's topic feel productive and forget efficiently.
Test yourself away from the tool. Close everything, write what you remember, then compare. The gap is your actual learning; everything you could only produce with the screen open is still borrowed.
And verify the content. A tutor that is confidently wrong about a formula, a date or a legal rule teaches the error very effectively. For anything you will be examined or judged on, check against your syllabus or a textbook.
Write it from memory, with everything closed. That is the only honest test.