It’s 10 PM in Lisbon. I’m on my friend’s couch, faced with a choice. I should be opening Anki, doing the mental reps that science has proven will build lasting knowledge. It’s my scheduled workout.
Instead, I open Bumble. The dating app demands nothing and promises everything, the cognitive equivalent of sugary sports drinks with quick satisfaction but minimal nourishment. Anki feels heavy, like a cold barbell promising strength I can’t yet feel, demanding immediate exertion.
From the outside, the two actions look identical: someone staring silently into a phone. But one is passive consumption; the other is deep cognitive strain, forging new neural pathways in quiet intensity.
You can’t tell.
This invisibility is the central challenge of knowledge work. We recognize physical effort—the sweat, the strain, the deliberate movement. But mental effort is silent. And AI, with its promise of effortless ease, makes this invisible struggle even easier to avoid. It floods our cognitive market with cheap conveniences, and our brains, naturally wired to conserve energy, will always choose the path of least resistance.
My decision to swipe instead of study wasn't a failure of willpower. It was a failure of design. It was a predictable outcome of a poorly structured cognitive environment.
The solution isn't to "be more disciplined." It’s to become a better architect.
Principle 1) Resistance is the Point
We go to the gym to seek resistance. Stressing a muscle makes it grow. The mind is no different. Effortful retrieval—actively recalling information just as you’re about to forget it—is what builds durable knowledge.
Cognitive scientists call this "desirable difficulty." Passive reading or watching a video summary is the mental equivalent of watching someone else lift weights. It creates the Illusion of Fluency, making you feel productive while building no actual strength.
A 2023 study of medical students found a direct, dose-dependent link between their use of digital flashcards—which automate this spaced retrieval—and their final exam grades. Forcing the brain to reconstruct a memory strengthens the neural pathways to that memory. The more effortful the rep, the stronger the recall.
When we ask an AI to "summarize this article," we are explicitly asking it to do our workout for us. We are outsourcing the very resistance that creates learning. The approach of an athlete, not a spectator, is to use AI to scaffold that resistance.
Don’t ask for a summary. Instead, after reading an article, make the AI your sparring partner.
Prompt: "I've just read [article/book chapter]. I will now attempt to (1) summarize the core thesis in one sentence, (2) explain the primary mechanism of action in my own words, and (3) propose one novel application of this concept to [my field, e.g., user onboarding for a SaaS product].
Your role is to act as a 'Red Team' critic. Do not affirm my understanding. Your only job is to attack my reasoning. Use questions like:
'What is the strongest counter-argument to your thesis summary?'
'Your explanation of the mechanism ignores what crucial variable?'
'Your proposed application has what potential second-order consequence you haven't considered?'
Let's begin. [Paste article text]."
You perform the reps. The AI makes them harder. That is how you get stronger.
Principle 2) You Must See Your Form
In the gym, bad form wastes effort and risks injury. In knowledge work, "bad form" is sloppy reasoning—unstated assumptions, logical fallacies, or imprecise communication. It’s the invisible source of most failed projects and confused teams.
Using AI to summarize a 32-email chain isn't improving your form; it's asking the trainer to lift a different, lighter weight for you. It bypasses the work. True "good form" means improving the clarity and structure of your own thinking. The AI's job is not to think for you, but to act as a mirror so you can see your own form clearly.
Before sending a high-stakes proposal or a critical email, make your AI a cognitive mirror.
Prompt: "Analyze the following text. Identify any assumptions I've left unstated, any claims that lack concrete evidence, and any instances of ambiguous language that could be misinterpreted. Present your feedback as a table with three columns: 'Claim,' 'Potential Weakness,' and 'Suggested Rephrasing for Clarity.' Do not rewrite it for me. Your job is to be the mirror for my thinking, not the author."
This is the equivalent of a coach correcting your posture before a heavy lift, ensuring your effort is applied with maximum effect and minimal risk.
Principle 3) Systematize Your Cognitive Rhythms
Athletes don't train continuously; growth happens during recovery. Minds require the same rhythm. The problem in Lisbon wasn't choosing leisure; it was the guilt-ridden, in-the-moment negotiation with myself. Willpower is a finite resource that is easily depleted by decision fatigue.
The solution is not more willpower, but better systems. It’s not believing blindly in the growth mindset and that you can do anything; it's having the wisdom to design a system that accounts for your human fallibility. Use AI as an executive function scaffold to build and enforce these systems.
Design your off-ramp from the workday.
Prompt: "Create a 10-minute 'shutdown ritual' for the end of my workday. The goal is to fully disengage from work to reduce cognitive residue. It must include three actions: (1) A task-capture step for tomorrow's top priority, (2) a prompt to identify one thing that went well today to prime for positive recall, and (3) a non-work transition activity (e.g., a 5-minute stretch, queuing a specific podcast). Generate this as a simple checklist I can run at 6 PM daily."
This automates the boundary between work and recovery. It removes the negotiation.
The prize for all this invisible, grueling work is not a higher IQ. It is cognitive sovereignty.
It is the agency to direct your own attention in a world engineered to hijack it. It’s possessing a mind that is not just a consumer of AI-generated summaries, but a producer of novel insights. Your AI is a partner. Like any partner, it can make you stronger, or it can make you weaker. The choice is not in the tool, but in the architecture of your engagement with it.
So, the real work is this:
This week, identify one routine task you've outsourced to AI. Redesign it. Re-introduce desirable difficulty. Make the machine your sparring partner, not your surrogate.
That is the only rep that counts.
what a brilliant analogy!
It all comes down to how you use the tool. You can certainly use it in a way where you offload learning. But you can also find ways to make it work more for you cognitively, not less if you choose to.