The Metacognition Method: How to Know What You Actually Know in the Age of AI (Actually Learn)

You understood it with the AI window open. Could you do it again tomorrow, alone, with nothing to look at?

Most people cannot say, and that is the problem this book is about. The feeling of knowing is not a readout of what is in your head. It is a guess your mind makes from cues: the answer sitting on the screen, the smooth explanation, the thing you saw an hour ago. The guess is often right. When it is wrong, it is wrong with full confidence, and you find out on the exam, in the meeting or in front of the client.

AI did not invent this problem, but it has made those cues cheap and constant. In one field experiment with nearly 1,000 high-school math students, those who practiced with a standard AI chat interface scored 48 percent higher during practice and 17 percent lower on a later exam taken without AI, compared with students who never had access. They did not perceive the drop. A version of the tool that gave hints instead of answers largely removed the harm. The lesson is not that you should avoid AI. It is that the tool quietly removes the signal that tells you what you have not learned yet.

The Metacognition Method rebuilds that signal. It turns metacognition, the skill of judging what you know and deciding what to do about it, into a routine called the Calibration Loop. One run takes about twenty minutes, on any subject:

• Go Cold: close the source, turn what you studied into questions with no answers in view, and wait.
• Place the Bet: next to each question, write 10, 30, 50, 70 or 90 percent for how likely you are to get it right.
• Prove It: answer every question cold, in writing, specifically enough to be wrong.
• Score the Gap: check each answer against an independent source and place it in the Knowing Grid, where the box that matters most is the Blind Spot: the things you would have bet on and got wrong.
• Route the Work: send each miss to the method that addresses it, and recheck in seven days.

You will learn how to:

• tell the difference between understanding something and merely recognizing it;
• spot secondhand certainty, the confidence that belongs to a source but feels like yours, after any AI-assisted session;
• put a number on how sure you are, and check it against a score you produced yourself;
• find your Blind Spots while there is still time to do something about them;
• stop restudying what you have already proved and spend the time where the gaps are;
• understand the Dunning–Kruger effect accurately, critiques included;
• use AI under five rules that keep your own judgment in the loop.

Every study in these pages is cited at its real strength, with its sample and its design, from pilots flying by feel in 1954 to a horse named Clever Hans, the Hubble Space Telescope and recent experiments on AI. Where a step is the book's own rule of thumb rather than a lab finding, the book says so. It is not an argument that AI is making anyone stupid, and it is not a tour of AI products. It is a study method for one adult learner at a time.

Written for professionals and serious learners who use AI every week, for anyone preparing for a certification, a degree exam or the domain of a new job, and for readers of the Actually Learn library who want to know which method a topic actually needs.

The Metacognition Method is Book 8 of the Actually Learn series by Gabriel Anhaia, alongside Zettelkasten, Reading That Sticks, Spaced Repetition, Deep Focus, The Feynman Method, The Active Recall Method and The Deliberate Practice Method.