The honest answer is that you already know how to stop. The problem is not knowledge — it is friction. AI removes the friction from learning, and friction is precisely where learning lives. Every time you ask a chatbot to explain something instead of wrestling with it yourself, you are trading a moment of discomfort for the illusion of understanding. You walk away feeling like you learned something, but the knowledge did not stick because your brain never had to work for it.
Daniel Kahneman's research on cognitive ease and cognitive strain explains why this matters. When information comes easily — when the font is clear, the explanation is smooth, the answer is instant — your mind enters a state of cognitive ease. You feel confident and comfortable, but you are also less critical, less engaged, and less likely to remember. It is cognitive strain — the state where things feel difficult and you have to slow down and think — that activates deeper processing. The struggle is not an obstacle to learning. It is the mechanism of learning.
So the first practical step is to reintroduce struggle deliberately. When you encounter a concept you do not understand, sit with it. Read the textbook passage again. Try to explain it to yourself in your own words before reaching for any tool. Give yourself at least twenty minutes of genuine effort before seeking help. This is not about suffering for its own sake — it is about giving your brain the resistance it needs to build actual neural pathways around the material.
The second step is to use the technique of teaching. After you study a topic, try explaining it to someone else — a friend, a study partner, even an empty room. If you cannot explain it clearly without notes, you do not actually understand it. This is far more revealing than any AI-generated summary, because it forces you to confront the gaps in your comprehension rather than papering over them.
There is a deeper issue here too, one that connects to what Brad Stulberg calls the mastery mindset. One of its core principles is to focus on the process rather than the outcome. When you rely on AI, you are optimizing for the outcome — getting the assignment done, having the answer ready. But when you focus on the process of learning itself — the reading, the confusion, the gradual clarification — you shift from obsessive achievement to genuine growth. The paradox is that people who focus on getting better actually end up performing better than people who focus on being the best.
A practical boundary that works for many people: use AI only after you have attempted the work yourself. Write your first draft before asking for feedback. Solve the problem incorrectly before looking at the solution. Form your own opinion before seeking another one. This way, AI becomes a tool for refinement rather than a substitute for thinking. The difference is enormous. One builds competence; the other erodes it.
Dorie Clark writes about the importance of doing hard things during your strong hours — not your weak ones. Apply this to studying. Do your most demanding cognitive work when your energy is highest, when you can tolerate the discomfort of not knowing. Save the easier review and organization for when you are tired. If you reach for AI every time something feels hard, you are training your brain that difficulty is a problem to be outsourced rather than a signal that real learning is happening.
The goal is not to never use AI. It is to use it the way a skilled carpenter uses a power tool — for specific tasks where it genuinely helps, not as a replacement for knowing how to build things with your own hands.