I was a terrible self-teacher for most of my life.
I'd buy the course, watch three videos, highlight a book into uselessness, and absorb almost nothing. I confused consuming material with learning it. They are not the same thing, and I had years of half-finished tutorials to prove it.
Then I changed one thing. I stopped using AI as a fancy search engine and started using it as a tutor — one that adapts to me, never gets bored of my dumb questions, and quizzes me when I'd rather quit. My learning speed roughly doubled. Not because AI is magic, but because it forced me to do the one thing I'd always avoided.
The fast way to learn with AI isn't asking it to explain things. It's asking it to test you. Passive reading feels productive and teaches almost nothing. Active recall — being forced to retrieve an answer — is what actually builds knowledge. AI is the perfect tool for active recall because it'll quiz you endlessly, explain at any level, and never judge the question. Use it to interrogate yourself, not to lecture you.
Here's the uncomfortable truth I had to face: reading and re-reading is one of the worst ways to learn, and it's the one almost everyone uses because it feels good.
It feels productive. You see the words, you nod along, you highlight. Your brain registers "I've seen this before" and mistakes familiarity for understanding. Then the test comes — or worse, real life comes — and the knowledge isn't there.
What actually works is retrieval. Closing the book and forcing your brain to pull the answer out. It feels harder. It feels worse. That difficulty is the point — the struggle is the learning happening.
For years I avoided the hard, effective method and clung to the easy, useless one. AI is what finally got me to switch — because it made the hard method frictionless.
The shift was simple. I stopped asking AI to tell me things and started asking it to quiz me on them.
After reading a chapter or watching a lesson, I do something like this:
"I just learned about [topic]. Ask me five questions, one at a time, from basic to hard. Don't show the answers. Wait for my response, then tell me what I got wrong and why."
Then I have to actually answer. Out loud or typed, no peeking. The AI catches the gaps, explains exactly the part I fumbled, and moves on. It's the experience of having a patient tutor on call, except it never sighs at me.
This one prompt pattern changed everything. I went from passively absorbing to actively struggling, which is where real learning lives. It's a close cousin of the prompt pattern that fixed my AI output — once you stop asking the model to perform and start asking it to interrogate you, the whole tool gets more useful.
Over time I built a small toolkit. These are the moves I reach for constantly.
I stopped asking AI to be smart for me. I started asking it to make me prove I'm not faking it.
A while back I needed to understand a technical area I knew nothing about, fast. The old me would have bought a course and ghosted it by lesson four.
Instead, I ran the loop. I asked for a plain-language overview, then immediately asked to be quizzed on it. I got things wrong. The AI explained why, in the exact spot I was confused. I asked it to connect each new idea to something I already knew. Then I tried to teach the whole thing back to it in my own words, and it flagged every place I'd gotten fuzzy.
Three focused sessions and I had a working grasp — not expert, but genuinely usable. The difference wasn't speed of reading. It was that I never let myself drift into passive mode. Every minute, the AI was making me retrieve, not just receive.
There's a failure mode here, and I fell into it early. It's letting the AI do the thinking and feeling like you learned something.
If you ask AI to summarize a book, read the summary, and feel accomplished — you've learned almost nothing. You've outsourced the exact mental work that constitutes learning. It's the highlighter problem with extra steps.
The rule I follow: AI should make me work harder, not less. If a session felt easy and comfortable, I probably wasted it. If I was straining to recall, getting things wrong, and being corrected — that's the productive discomfort. Lean into the part that feels bad.
The same goes for AI-powered note-taking and study tools. They're great for organizing and quizzing. They're useless if you let them think instead of you. The model is the gym, not the bus. It's there to make you lift, not to carry you. This is exactly the trap I keep returning to in the honest truth about AI productivity tools: the tools that feel effortless are usually the ones quietly doing your thinking for you. Decades of learning-science write-ups, including the ones summarized by Harvard Business Review, keep confirming that retrieval beats re-reading by a wide margin.
The single quiz session is powerful. A repeatable loop is what compounds. Here's the rhythm I settled into, and why each piece earns its place.
I learn in small chunks, not marathon sessions. Twenty to thirty minutes of new material, then I close it. Your brain consolidates knowledge between sessions and during sleep, so three short sessions across a few days beat one long cram that leaks out by morning. Spacing isn't a productivity hack; it's how memory physically works.
After each chunk, I run the recall loop immediately while it's fresh — quiz me, explain my mistakes, connect it to what I know. Then, and this is the part most people skip, I come back the next day and ask to be quizzed on yesterday's material cold, before reviewing it. The forgetting-then-retrieving is exactly where the durable learning happens. It feels uncomfortable because I've half-forgotten it. That discomfort is the signal that I'm building something real, not just refreshing a recent memory.
Once a week, I ask the AI to give me a mixed quiz across everything I've covered. This is interleaving — mixing topics instead of drilling one at a time — and it's one of the most reliable ways to make knowledge stick and transfer. It's harder than blocked practice and it works far better.
The whole loop takes a fraction of the time my old highlight-and-reread habit did, and it actually deposits the knowledge somewhere I can reach it later. That's the difference between feeling like you learned and actually having learned.
If there's something you've been meaning to learn, try the quiz-me loop on it this week instead of another course you won't finish — twenty uncomfortable minutes will teach you more than an hour of nodding along.
Q: Won't AI just give me wrong information sometimes? Yes. It can be confidently wrong, especially on niche facts. For anything that matters, verify against a real source. But for building understanding and getting quizzed, the occasional error matters less — and catching its mistakes is itself good practice.
Q: What's the best prompt to start with? "Quiz me on [topic], one question at a time, basic to advanced, and explain what I get wrong." That single loop is responsible for most of my gains. Start there before anything fancier.
Q: Is this just for technical subjects? No. It works for languages, history, business concepts, anything you can be questioned on. Anything with right-and-wrong answers or arguments to reconstruct is fair game for the recall loop.
Q: How long should a session be? Short and frequent beats long and rare. Twenty focused minutes of getting quizzed beats two passive hours. Your brain consolidates between sessions, so spacing them out helps the knowledge stick.
I didn't get faster at learning because AI handed me answers. I got faster because it finally made me do the hard thing I'd avoided my whole life: closing the book and forcing the answer out of my own head.
AI is a phenomenal tutor and a terrible crutch. The difference is entirely in how you use it.
Don't ask AI to know things for you. Ask it to make you prove you know them. That discomfort is the whole point.
Pick something you've been "meaning to learn" and try the quiz loop today. Twenty minutes. Get things wrong on purpose. You'll learn more than from the last three courses you didn't finish.
Thirty books, hundreds of highlights, and one uncomfortable realization: the handful of ideas that actually mattered would fit on an index c…
I spent years saving the hardest task for when I 'felt ready.' Doing it first instead quietly fixed my focus, my dread, and my output.
I tracked every distraction for a week and was horrified by what I found. Then I fixed the three that mattered most.
Comments
Sign in to join the conversation
No comments yet. Be the first to share your thoughts!