The Effort Is the Point

I was stuck on a problem, one of those knotted questions where every path forward seemed to contradict every other path. I'd been circling it, jotting down notes that led nowhere.

So I opened Claude and started arguing with it.

Not querying. Arguing. I'd make a claim, it would push back, I'd refine, it would find the hole, I'd patch the hole, it would find another. An hour in, I'd progressed more than I had in the previous two days, not because the AI solved it for me, but because having something to push against made my own thinking faster, sharper, more precise.

I sat back and looked at what we'd produced. I couldn't have done it alone. Not because I lacked the capability, but because I lacked the interlocutor, the friction of another mind (or something like a mind) that wouldn't let me get away with my usual shortcuts.

That's when I realized I'd been thinking about this wrong.

The anxiety goes like this: AI will do what we do, but faster and cheaper, until there's nothing left for us to do. We'll atrophy. Our skills will wither. We'll become supervisors of processes we no longer understand, then not even that. Somewhere between "tool" and "replacement," we'll lose the thing that made us us.

I've felt this anxiety in my body. I've caught myself reaching for the AI to finish a sentence and wondered if I was training myself into helplessness. I've watched my own fingers hesitate over the keyboard, unsure whether the effort of thinking was still worth it when the shortcut was right there.

The fear is real. I won't pretend otherwise.

But I've also started to notice something else, a counter-signal, equally real, that the anxiety tends to obscure. When I work well with AI, I don't feel diminished. I feel like I'm thinking at a higher level than I could alone. It is not because the machine is doing my thinking, but because it's holding things for me, context, connections, half-formed possibilities, so that I can focus on the part that actually requires me. The judgment. The taste. The decision about what matters.

It's not unlike what happens with any powerful tool. A carpenter with a nail gun can build faster than one with a hammer, but speed isn't the point. The point is that the nail gun frees attention for the craft, the angles, the joints, the feel of the wood. The tool doesn't replace skill-it relocates it.

The question is whether AI relocates skill or dissolves it. And I don't think we know yet. I don't think it's determined yet. We're in the middle of something, and what we do now shapes what it becomes.

Hegel is a philosopher I keep returning to. He was a 19th-century German who never saw a computer but understood something about how change actually works. His insight was that change isn't smooth. It happens when contradictions within a system become unbearable and tear it apart. But the tearing isn't pure destruction. Something is cancelled, something is preserved, something is elevated into a form that transcends the original opposition.

Hegel called this Aufhebung, in English, Sublation, a word that means, impossibly, all three at once. To destroy, to preserve, to raise up.

The caterpillar is destroyed in the cocoon; something of the caterpillar is preserved; what emerges can fly.

When I look at what's happening to knowledge work, I see Sublation in progress. The contradiction was always there: we called ourselves "knowledge workers" but spent most of our time on non-knowledge work. Formatting spreadsheets. Chasing emails. Building decks few would read. The job was sold as thinking, but often experienced as bureaucratic friction.

AI is dissolving that friction. Which means the contradiction is being forced into the open. The fake parts of the job, the parts that were always just overhead, are getting stripped away. What remains is either the real thing or nothing at all.

That's terrifying if what you had was mostly overhead.

It's liberating if what you have is something real.

The question that haunts me is the meaning question.

If AI can do the work, why should I bother doing it myself? What's the argument for the slow, hard thing when the fast, easy thing is indistinguishable?

I used to think the answer was quality, that human work was somehow better. But that's increasingly untrue, or at least increasingly complicated. The AI's output is often fine. Sometimes better than fine.

So I've had to find a different answer, and here's what I've got:

The meaning was never in the output. The meaning was in the doing.

When I write a sentence myself—when I feel my way through the options, choose a word, reject it, choose another—something happens in me that doesn't happen when I approve a sentence the machine wrote. Even if the sentences are identical. Even if no reader could tell the difference.

The experience is different. And experience is where I live. It is the same when cooking a meal from scratch versus ordering out, or savoring a vegetable from your own garden instead of a can.

This isn't an argument against using AI. It's an argument for noticing, for paying attention to which activities are valuable because of what they produce, and which are valuable because of what they do to me while I'm doing them.

Some things I'm happy to hand off. The overhead, the drudgery, the stuff that was always just in the way. Take it. Please!

But some things I'm keeping. Not because I'm better at them than the machine, but because doing them is who I am. Because the effort is the point. Because I'm not optimizing for output, I'm optimizing for a life that feels like mine.

I want to be honest about the hard part. The juniors. The entry-level workers who used to learn by doing the drudgery, who built their skills through repetition, who became seniors by first being bad at the job for several years until the pattern recognition kicked in. AI does junior work. That's what it's capable of for the time being: first drafts, simple analysis, boilerplate code, routine research. Companies are making the obvious calculation: Why hire a human to do what the model does better?

So the entry-level roles are thinning. The ladder is losing its bottom rungs. And there's a real question, one we don't have a clear answer to, about how the next generation develops expertise when the apprenticeship has been automated away. As the father of two who have yet to enter the workforce, this is troubling.

But I've also noticed the problem is visible now. People are talking about it. Companies that care about their future are experimenting with new models, AI-assisted training, and hybrid roles where juniors learn by directing AI rather than doing the tasks themselves.

Is that sufficient? I don't know. It might not be. But it's not nothing. And the historical pattern, for what it's worth, is that these problems get solved, not automatically or without struggle, but through the ingenuity of people who refuse to accept that the first version of the future is the final one.

The Luddites weren't wrong that the power loom would destroy their livelihoods. They were wrong that weaving would cease to exist. I'm not saying history guarantees a happy ending with AI. I'm saying it doesn't guarantee an unhappy one either. The synthesis isn't written yet. We're the ones writing it.

Here's where I've landed, at least for now.

The atrophy fear is real but incomplete. Yes, if you use AI to avoid thinking, you'll get worse at thinking. If you outsource every sentence, your sentences will weaken. The muscle atrophies when you stop using it.

But that's true of any tool. Calculators didn't destroy mathematical intuition; they destroyed it in people who stopped doing math. They enhanced it in people who used the calculator to solve problems they couldn't have touched otherwise. Same tool, different relationship, different outcome.

What matters isn't whether you use AI. It's how you use it, and why, and what you're trying to become.

I use AI to argue with. To stress-test my thinking. To generate options I can react to, because reacting is often faster than generating from scratch, and the output is still my responsibility. I use it to hold complexity so I can focus on judgment. I use it to find the edges of what I know, which is where the interesting work begins.

And crucially, I use it for things that aren't the point, so I can spend myself on things that are. The formatting, the first drafts, the research synthesis, I've let a lot of that go. Not because I couldn't do it, but because doing it was never where my value lived. It was overhead. It was the tax I paid to get to the real work. And now that tax is lower, which means I can spend more time on real work.

That's allocation, not atrophy. I'm not becoming less capable. I'm becoming more focused. The capabilities I'm losing are the ones I never cared for in the first place, the ones that were always in the way of what I actually came here to do.

The synthesis is coming whether we like it or not. That's what Hegel understood: the dialectic doesn't wait for permission. The contradiction was there; now it's resolving.

But which synthesis we get, that's still open. That's still being written, by a million choices, by policy and culture, and individual decisions about how we relate to these tools.

My career is built on the bet that this technology makes things better. And I still believe that, not naively, not without anxiety, but genuinely. Because I've seen what it looks like when it works. I've felt my own thinking sharpen against the resistance of a machine that won't let me be lazy. I've watched people do things they couldn't have done before, reach problems they couldn't have touched, become more capable rather than less.

The dystopia is possible. I won't pretend otherwise. A world of atrophied skills and pulled-up ladders, where the gains flow to the top, and everyone else scrambles for the scraps, I can see how we get there from here.

But I can also see the other path. The one where AI handles the parts of work that were never really the point, and humans get to focus on the parts that are. The one where the entry-level problem gets solved, not automatically, but because people who care about the future build new ways to develop expertise. The one where we come out of this transition more capable than we went in, not less, because the tools revealed what we could become when we stopped spending ourselves on overhead.

That path requires choices. It requires politics and power and struggle, same as every previous transition. Nothing's guaranteed.

But I don't think it's naïve to work toward it. I don't think it's denial to believe it's possible. I think it's the job. The real job, the one that was always hiding under all that overhead.

The synthesis isn't something that happens to us. It's something we make.

I'm sitting here with my hands on the keyboard, and I have a choice. I can let the machine finish this thought, or I can finish it myself.

Both options are available. That's new. That's the condition we live in now.

And what I've decided, for this sentence, for this moment, is that I want to feel my way through it. Not because the output will be better. But because the feeling is mine. Because the choice is mine. Because I get to decide what parts of this strange new world I embrace and what parts I hold back from.

The machine can finish my sentence. That's the thesis.

But I can also finish it myself, in collaboration with the machine, in ways I couldn't have imagined before. That's the antithesis.

And somewhere between those two, in the tension, in the choosing, in the daily negotiation of what to hand off and what to hold onto, the synthesis is emerging.

I don't know exactly what it looks like yet.

But I know I want to be one of the people who shape it.