Think about the last time you were genuinely stuck.
Not frustrated stuck or impatient stuck. Genuinely, productively stuck, holding a problem that resisted your first three approaches, that wouldn’t resolve into the frameworks you already had, that required you to sit in the discomfort of not-yet-knowing long enough for something new to arrive.
That experience has a specific texture. A particular quality of attention that is simultaneously uncomfortable and alive. Where the mind is working without producing, turning the problem, approaching it from different angles, discarding partial solutions, holding the tension between what you know and what you don’t yet know. It feels inefficient. It feels slow. From the outside it looks like nothing is happening.
Inside it, everything is happening.
That experience is thinking. Not the only form of thinking. But the form that matters most. The form that produces genuine understanding rather than processed information. The form that changes the thinker rather than just delivering an answer to the thinker.
And it is precisely this experience, this specific, uncomfortable, generative stuck-ness, that AI makes unnecessary.
Which sounds like a feature. It is also a cost. And the cost is almost never named clearly enough to be examined.
When you hand a question to AI before sitting with it yourself, you are not just getting the answer faster. You are skipping a process that was doing something you cannot see because it was doing it inside you.
Here is what that process was doing.
It was building a structure. Every genuine thinking encounter, every moment of holding not-knowing, of approaching the problem from a new angle, of discarding what doesn’t work and following what does, deposits something in the architecture of your understanding. Not information. Structure. The internal organization of what knowing allows you to navigate similar problems in the future not by remembering the answer but by having developed the judgment to find it.
This is what Fernando Flores meant by background. And what Heidegger meant by the ready-to-hand. The expert doesn’t consciously recall solutions, they have developed a structural readiness, built through thousands of encounters with problems, that makes the right move available without deliberate retrieval. The background does the work. Invisibly. Instantly. In the way that only accumulated experience can.
That background is built through the struggle. Through the stuck-ness. Through the process of working through problems without shortcuts. Every time you hand the problem to AI before the struggle has a chance to do its work, the background doesn’t get built. The structure doesn’t develop. The judgment that would have emerged from this encounter doesn’t emerge.
You got the answer. You didn’t become the person who could find the answer.
Now multiply that by every question you’ve handed over this week. This month. This year.
The accumulation is invisible because each individual substitution is so reasonable. This question isn’t worth the struggle. I’m busy. The output is good enough. I’ll engage more deeply next time. Each decision is completely defensible. The cumulative effect, the background that isn’t being built, the judgment that isn’t developing, the structural readiness that isn’t accumulating, is completely invisible until it suddenly isn’t.
The programmer who can generate code but can’t debug it. The writer who can produce fluent prose but can’t find the argument underneath it. The analyst who can prompt for insights but can’t evaluate whether they’re right. The student who can produce the essay but hasn’t developed the thinking the essay was supposed to build.
Each one is not a failure of intelligence. Each one is the entirely predictable consequence of outsourcing the formation process before the formation was complete.
Here is the precise thing you are outsourcing when you hand a question to AI before sitting with it.
Not the answer. The encounter with your own not-knowing.
And that encounter, uncomfortable, inefficient, resistant to shortcuts, is the most valuable cognitive experience available to you. Because it is the only experience that builds the thing that cannot be built any other way. The background. The judgment. The structural knowing that is yours rather than borrowed. That lives in you rather than in the tool. That is available when the tool isn’t, when the tool is wrong, when the situation requires judgment that no tool can provide because it requires the background that only you could have built.
This is not an argument for doing everything the hard way. A calculator doesn’t prevent you from developing mathematical intuition if you’ve already developed it. A GPS doesn’t prevent you from developing spatial reasoning if you already have it. The tool in service of formed capacity is one thing. The tool substituting for the formation of capacity is something else entirely.
The question is not whether to use AI. The question is whether the capacity that the AI is serving has been formed. Whether the background is there. Whether the judgment is developed. Whether you are using the tool to extend your thinking or using the tool to avoid the formation of thinking.
And that question requires knowing what thinking is. What it requires. What it builds. What the struggle is actually doing when it feels like nothing is happening.
That’s where this series is going.
Because the tool is not the problem. The unexamined relationship to the tool is the problem.
And you cannot examine a relationship you don’t have language for.
You’re building the language now.
@ Lee Greene