There is a difference between having an answer and arriving at one.

That distinction might sound minor on the surface, the way a lot of important things sound minor until they don’t. It sits at the center of something most people are not yet willing to say out loud about artificial intelligence: that the threat is not that it will think incorrectly. The threat is that it will think so correctly, so quickly, and so effortlessly, that we will stop being the kind of people who think at all.

A study published earlier this year tested more than twelve hundred participants on math reasoning and reading comprehension tasks. One group received AI assistance during the task. The other did not. When the AI-assisted group returned to working on their own, without any help, they performed measurably worse than they had before the study began, and they were significantly more likely to give up when problems became hard. The researchers were not measuring the effect of months of AI dependency, or years of it. They were measuring the effect of a single, brief interaction. Ten minutes. The erosion started there.

That finding confirms something I have been watching in the men I work with, something I could not yet point to with data but recognized in the field of every conversation. There is a quality of presence that disappears first. Then the willingness to sit inside a hard question without reaching for a rescue. Then, finally, the confidence that their own mind is even worth trusting.

This is not a technology problem. It is an identity problem.

In the work I do inside Resonant Identity Theory, I describe eleven states through which a person’s sense of self either expands toward freedom or contracts toward suppression. The Suppressed state is not characterized by dramatic collapse. It does not look like crisis. It is characterized by the slow, almost comfortable delegation of the self to whatever external structure promises relief from the experience of uncertainty. It can be a relationship. It can be religion. It can be a job, a denomination, a movement, an ideology. Now it can be a prompt box and a blinking cursor.

What researchers have begun to call “borrowed certainty” is the invisible contract of the Suppressed state. Borrowed certainty is the feeling of understanding that arrives without requiring the work that understanding has always demanded. You stop thinking in exchange for the comfort of having answers. The contract looks like efficiency. It feels like intelligence. And it costs you the very architecture of a self.

Freedom, the kind I call the Unbound state, the place every framework I build is oriented toward, is not the absence of difficulty. It is the demonstrated capacity to meet difficulty and remain yourself inside it. You cannot shortcut to that. You cannot borrow it. You cannot prompt it into existence.

Anders Ericsson, the cognitive psychologist whose research gave us the scientific foundation for what Malcolm Gladwell later popularized as the ten-thousand-hour rule, showed through decades of study that expertise is not primarily a function of natural talent. It is a function of sustained, deliberate engagement with difficulty. Not practice in the casual sense. Deliberate practice. The kind that consistently places you at the edge of what you can currently do and keeps you there until the edge moves. Ten thousand hours is not a number about time. It is a number about the willingness to stay inside discomfort long enough for something real to be formed.

What the study referenced above reveals is that ten minutes of outsourced cognition begins to erode the very psychological precondition for that willingness. Not the skill. The appetite. The tolerance. The orientation of the self toward the difficult thing.

That is a different, and far more serious, kind of loss.

A related phenomenon has shown up in medical research as well. Gastroenterologists who used AI assistance to detect polyps saw their detection rates improve while the technology was running. When the technology was removed, their detection rates dropped below where they had started before using AI at all. Not a return to the baseline. Below it. Researchers studying pilots, drivers, and writers have found similar patterns. The skill does not simply pause while a tool takes over. The skill begins to recede. The cognitive circuitry that was being exercised, because it is no longer being exercised, weakens. The ten-minute study suggests that the clock on that degradation starts far earlier than anyone assumed.

That is not abstract. It is what happens to a man who has been operating from an external reference point for long enough that his own internal compass no longer reads true. I have watched men walk out of twenty-year institutional systems, systems that told them what to believe, what to feel, what to want, and they could not locate themselves. Not because they lacked intelligence. Because they had delegated their cognition so thoroughly, and for so long, that recovery required something closer to excavation than education.

The AI version of this is faster. That is the terrifying part.

There is a further dimension to this problem, one that tends to get overlooked in conversations that frame AI dependency primarily as a productivity or performance issue. Human cognition is effortful and time-dependent. We earn fluency after struggle, and the confidence that follows has a path we can retrace. That retraceable path is not just about epistemology, about how we know things. It is ontological. It is about who we become in the knowing of them.

Every man I have ever guided who arrived at what I recognize as genuine freedom could trace the line between who he was before and who he became. The line always runs through a struggle he did not escape. Through something he sat inside long enough to be changed by it.

What concerns me about the world this research describes is not that people are using AI. I use it. The question is what posture they carry into that use, and more importantly, what posture they carry out of it.

There is a meaningful difference between using AI as a provocation and using it as a resolution. The person who uses a tool to push deeper into their own thinking is still sovereign over their mind. They arrive with something already forming. They bring a half-built idea to the conversation and use the exchange to test it, sharpen it, challenge it. The person who opens the tool to avoid the feeling of not knowing yet has handed something over that they may not realize they lost until the day they reach for themselves and find nothing there.

The study frames AI dependency primarily as a cognitive problem. A problem of persistence, of performance, of measurable skill degradation. Those concerns are real and well documented. But the deeper problem is not that people are losing a cognitive capacity. It is that they are losing the experience of being the one who thinks, and in losing that experience, they are losing a constitutive layer of self.

Your thoughts are not incidental to who you are. They are where you discover who you are. The moments of highest personal formation in my own life, and in the lives of the men I have sat with, almost never arrived in smooth water. They arrived in the middle of struggle, in the space where the answer had not yet come and something in me was forced to remain with what I did not yet know. That remaining, that inhabiting of the unresolved, is where identity is made.

When AI short-circuits that space, it does not merely rob a person of a cognitive exercise. It robs them of the raw material of self-formation.

This is the invisible contract nobody reads before signing it. Every time a person prompts a model to answer a question they could have wrestled with themselves, they are trading away a tiny increment of their own cognitive authority. The answer feels like a gain. It is a loss. Not because AI answers are wrong. Often they are extraordinarily right. But rightness purchased without struggle does not become yours in any meaningful sense. It remains borrowed. Rented. Held at the address of someone else’s mind.

This is not a call to reject the tools. Csikszentmihalyi’s concept of flow, the state of optimal human engagement where challenge and capacity are in near-perfect alignment, lives on the edge of chaos, not in its absence. Maslow’s peak experiences come after the laborious climb. The architecture of human cognition requires something to push against. AI, used deliberately, can be that thing. Used by default, it removes it.

The question worth asking is not whether AI is capable. It clearly is. The question is whether we are maintaining the kind of selves for whom capability has meaning. Whether we are still people who believe that the journey toward understanding is not just instrumental, not just a means to get answers, but is itself the site of our formation.

I know this from working with men who have spent years inside systems that thought for them: the return to self is possible, but it is never quick, and it is never painless, and it requires the willingness to sit inside the experience of not knowing without reaching for a rescue.

That willingness is freedom. Not the freedom to have answers. The freedom to be the one who seeks them.

And no tool, however capable, however fluent, however confident in its completion, can be that for you.

The mind that never stumbles doesn’t grow. But worse than that, the man who never stumbles never learns that he was the one capable of standing back up.

That is the identity at stake here. And it is worth protecting.

8-Step Plan to Reduce AI-Assisted Thought Dependency

This is not an anti-technology plan. It is a pro-self plan. The goal is to restore cognitive sovereignty so that when you do use AI, you are directing a tool rather than being carried by one. Each step is ordered deliberately. Do not skip to the interesting ones.

Step 1: Audit Your Current AI Prompting Habits for 7 Days

Before you change anything, you need to know what you are actually doing. For one full week, log every time you open an AI tool. Write down what you were doing before you opened it, what question you asked, and whether you had made any attempt to answer it yourself first. No judgment. Just data. You cannot address a pattern you have not yet seen clearly enough to name.

Step 2: Establish a Pre-Prompt Friction Rule

Effective immediately, you do not prompt an AI model until you have spent a minimum of ten uninterrupted minutes with the question on your own. Not searching the web. Not texting someone. Ten minutes alone with a blank page, a pen, and the question. You write whatever you know, whatever you half-know, and whatever you are genuinely unsure of. Then, and only then, you open the tool. This ensures that AI functions as a conversation partner rather than a replacement thinker. You arrive with something already forming.

Step 3: Designate Two AI-Free Thinking Windows Per Day

Choose one window in the morning and one in the afternoon, each no shorter than thirty minutes, in which no AI tools are opened at all. During these windows, you think, write, read, or sit with whatever problem is in front of you. The goal is to rebuild the tolerance for cognitive friction that research now shows begins to erode after a single ten-minute session of AI dependency. Discomfort during these windows is not a sign that something is wrong. It is the sign that something is working.

Step 4: Return to Longhand Writing as a Primary Processing Mode

Your journal has been a cognitive anchor since high school. Lean back into it, not as a spiritual practice alone, but as a deliberate cognitive exercise. Longhand writing forces a different relationship with thought than typing does. It is slower, which means ideas must be held in mind longer before they become words. This delay is not inefficiency. It is the friction that builds depth. Commit to a minimum of 500 words per day written longhand before any screen is opened.

Step 5: Practice Deliberate Incompletion

Once per week, take a significant question, a framework problem, a teaching concept, a business decision, and commit to leaving it unresolved for 48 hours. No AI, no Google, no asking someone. Let it sit. The experience of living inside an unresolved question without collapsing toward a rescue is the core cognitive skill that dependency degrades. This step rebuilds it directly. You are not waiting for the answer. You are training the capacity to wait.

Step 6: Reconstruct AI-Assisted Outputs by Hand

For anything important that AI has helped you produce in the past, particularly frameworks, course structures, or written arguments, take one piece per week and reconstruct it from memory without looking at the original. Write it from scratch as if the AI output did not exist. What you cannot reconstruct without help is something you borrowed but never owned. This is not about distrust of the tool. It is about ensuring that what you produce through AI has been metabolized into genuine understanding.

Step 7: Establish a “Thinking Out Loud” Practice with a Person, Not a Prompt

Identify one person, a peer, a colleague, a trusted conversation partner, with whom you talk through complex problems verbally before seeking any outside answer source. This is not brainstorming for the sake of collaboration. This is the practice of articulating a half-formed thought in real time, which requires that you tolerate the exposure of not yet knowing. AI cannot give you that exposure. Only another human being who witnesses your uncertainty can. If you do not have that person in your life right now, this step becomes a requirement for building it.

Step 8: Monthly Cognitive Sovereignty Review

At the end of each month, return to the following four questions and write your honest answers. First: what did I figure out on my own this month that I would have previously outsourced? Second: where did I feel the pull toward AI strongest, and what was I actually avoiding when I felt it? Third: what questions am I currently willing to sit inside without needing an immediate answer? Fourth: is the quality of my own thinking, my original analysis, my novel frameworks, my unassisted writing, getting stronger or weaker? This review is not accountability in the institutional sense. It is reconnaissance. You are mapping the territory of your own mind and checking whether you are still the one who lives there.