Sometime around 1980, I was sitting in a library at Purdue at an hour no one should be in a library. I was in an advanced writing course, the kind where the reading was heavier than the writing, and I’d been grinding through an assignment for longer than I want to admit.
I don’t remember the specific paper. I remember the moment.
Hours into it, after cycling through sources that contradicted each other, after re-reading the same paragraphs and not understanding why I kept returning to them, something reorganized. The threads I’d been pulling on separately connected. Not neatly. Not like a puzzle clicking into place. More like a lens focusing. The material hadn’t changed. I had.
If you’ve had that experience, you know what it feels like. There’s a physical quality to it. A rush. Almost a kind of euphoria. Your mind has been doing slow, invisible work for hours, and then the structure appears all at once. You didn’t build it consciously. You just suddenly see it.
I carried that memory around for decades without quite knowing what to do with it. Then I remembered where I first found language for it.
Phaedrus in the Stacks
In “Zen and the Art of Motorcycle Maintenance,” Robert Pirsig tells the story of a man named Phaedrus who is studying philosophy at the University of Chicago. Phaedrus has become consumed by a single question: what is Quality? Not quality as a casual word. Quality as a philosophical concept that sits underneath everything we recognize as good or true or beautiful.
He can’t define it. Every definition he tries either collapses under scrutiny or reduces Quality to something smaller than what it actually is. His professors want a definition. His classmates want a definition. The Western philosophical tradition demands definitions. And he doesn’t have one.
So he does what students have always done when they’re stuck. He goes deeper into the stacks. He reads Plato. Aristotle. Kant. He follows threads for weeks. He’s exhausted, barely sleeping, approaching something that looks from the outside like obsession.
And then it breaks open.
Quality, he realizes, doesn’t sit on the subjective side or the objective side. It precedes both. It’s the event that creates the split between subject and object in the first place. And with that single insight, everything he’s been reading reorganizes. The entire history of philosophy looks different from that angle.
Pirsig writes the moment as electric. Not triumphant. Something closer to dangerous clarity. The kind of understanding that arrives only after the mind has exhausted every easier path and been forced to reorganize itself around something new.
I think that’s what happened to me in the Purdue library, in a smaller and less philosophical way. And I think most people who’ve done sustained intellectual work have a version of that story. A night when the confusion wasn’t an obstacle to understanding. It was understanding, in progress.
Years later, during a conversation in my weekly men’s group, I said something offhand that I hadn’t planned. “Nothing of great value was ever forged in the furnace of comfort.” The room went quiet for a second, and then someone wrote it on the wall. It’s still there.
I wasn’t thinking about Pirsig when I said it. But it’s the same idea. The furnace is the point. The discomfort, the confusion, the hours where nothing seems to be working. That’s not the price you pay for the breakthrough. That’s the process that produces it.
What AI Offers, and What It Costs
I build AI systems for a living now. I use Claude every day. I’ve built dashboards, workflows, a chatbot, a six-week cohort where I teach other people to build their own AI command centers. I am not writing this as someone who stands outside the technology and warns about it from a safe distance.
I’m writing it as someone who uses it constantly and keeps noticing something.
AI is extraordinarily good at compressing the early parts of intellectual work. The research. The synthesis. The first draft. The organization of scattered material into coherent structure. Tasks that used to take hours now take minutes. I teach people how to do this. I watch their faces when they realize what’s possible.
But I keep thinking about that library.
The hours I spent cycling through contradictory sources weren’t wasted time. They were the mechanism. My mind was doing something during those hours that it couldn’t have done any other way. It was holding contradictions without resolving them. It was building an internal model of the problem that no summary could have provided. It was failing, repeatedly, to understand, and in the process developing the capacity to understand at a level I hadn’t been capable of when I walked in.
The euphoria at the end is what a mind feels when it has reorganized itself around something new. That reorganization required the hours. It required the frustration. It required the confusion that felt, in the moment, like it was going nowhere.
AI would have handed me the synthesis in four minutes. It would have been useful. I would have moved on to the next thing. And the reorganization that happened in my brain that night would never have occurred.
The Tension I Can’t Resolve
I’m not making the argument that we should all go back to card catalogs and pencil notes. The practical gains from AI are real, and I help people capture them every week. When someone needs to understand a schema, summarize a report, draft a framework, or map a process, AI is the best tool most of us have ever had access to.
But there is a category of understanding that can only be earned through sustained cognitive struggle. The kind that happens when you sit with something long enough that your mind is forced to build new structure rather than retrieve existing structure. Pirsig’s Phaedrus didn’t need a faster research tool. He needed the weeks of grinding to break down his existing mental models so a new one could form.
And I don’t think we’re talking about this enough.
The conversation around AI and knowledge work has mostly been about efficiency. How much faster can you get to the answer. How many steps can you eliminate. How much of the grunt work can you hand off. Those are the right questions for a lot of tasks. They are the wrong questions for the kind of deep work that actually changes what a person is capable of thinking.
I keep coming back to a simple version of the concern. If AI saves us from the suffering, does it also save us from the understanding that only comes through it?
There’s a one-act play by Thornton Wilder called “The Angel That Troubled the Waters.” A physician comes to the healing pool at Bethesda, hoping to be cured of his melancholy. The angel appears, stirs the water, and blocks him. The physician begs. The angel refuses. And then the angel says something that, years later during a season I won’t get into here, I had tattooed on my forearm: “Without your wounds, where would your power be?”
The play makes the case that the physician’s suffering is the very thing that allows him to reach people that even angels can’t. His wounds aren’t a deficiency to be cured. They’re the source of everything he has to offer.
I think about that when I watch AI smooth the path for people. Faster answers. Cleaner drafts. Less friction. All genuinely useful. But some of what gets smoothed away is the thing that was building capacity in the first place.
I don’t have a clean answer. I hold both sides of this honestly. I built my career on helping people use AI effectively. I also remember what it felt like to sit in a library at an hour that made no sense, confused in a way that felt permanent, and come out the other side understanding something I couldn’t have understood any other way.
That moment can’t be compressed. I’m not sure it should be.