The Trap of Limitless Capability

By Nathan Waterhouse · 2026-02-17

It's Saturday. 4pm. I've just spent the last five hours in the woods. Deliberately. I'm refusing to go back inside.

If I go back inside, they'll be waiting for me. Claude Code. ChatGPT. Gemini. Terminal open, cursor blinking. I know what they can do. They could clear my entire to-do list in half an hour if I ran them all in parallel. That's the lure. That's the temptation. And that's the trap.

The irony isn't lost on me that I'm dictating this piece to Claude through a headset while walking through the trees to get its feedback. I'm talking to an AI about how AI is consuming my time. But that irony is part of the story.

The week I lost

I haven't been outside once this week until today. That's why I'm out here: five hours and counting, because the alternative was another evening swallowed by capability.

The latest models are extraordinary. They hallucinate less. They build with a sophistication that sometimes exceeds what I could do myself. Code, presentations, contracts, marketing campaigns. There's no obvious ceiling anymore.

And that's the problem.

Instead of freeing me up, these tools have made me busier than I've ever been. I'm working longer hours. I'm working harder. And the question I keep returning to is: what has actually changed?

I think my work is better. But real life outcomes? Those have got worse. I've been ill this week from stress. I haven't seen my kids and my wife enough. My health is deteriorating. The tools that promised to give me time back have taken more of it than ever before.

This isn't just me

I look at the AI communities I'm part of and I see the same thing everywhere. A Harvard Business Review study recently confirmed what many of us are feeling: early adopters of AI tools aren't working less. They're working more.

The pressure comes from two directions. From inside, the speed feeds your motivation — you finish something quickly, so you start something else. You finish that, so you start another. From outside, there's the competitive anxiety: your competitors have these tools too. If you slow down, they won't. Your job might become redundant. Your industry might shift underneath you.

Both push the same way. More. Faster. Again.

We've been here before

This pattern isn't new. Juliet Schor's research traces a line through every major productivity revolution in history, and it tells the same story each time.

The agricultural revolution should have given us abundance and leisure. Instead, it trapped us in monoculture: longer hours, worse nutrition, less variety than the foragers who came before us. The industrial revolution promised liberation through machinery. What it actually delivered, particularly in the mills of Northern England, was closer to slave labour. People worked harder, in worse conditions, for longer.

Each time we've built tools that should have freed us, we've used them to raise the baseline expectation instead. We chose more output, not more life.

And conditions only improved when people fought for it. When they made deliberate, collective choices about how those gains would be shared. It never happened on its own.

Yeah, but things got better, right?

There's a fair counter to all of this, and I want to take it seriously.

Peter Diamandis and others have argued convincingly that technology makes life better over time. Healthcare, communication, education — all transformed by waves of innovation. Kevin Kelly makes a related point: technology expands the choices available to us, and more choice is a good thing. By that logic, what I'm describing is a rough patch. The long-term trend bends positive. Relax.

They're probably not wrong. Things probably will get better.

But there's a pattern hidden inside that progress. Craig Lambert wrote about what he called "shadow work" — the invisible labour that technology quietly shifts onto the rest of us. Self-checkout. Online booking. Managing your own tax returns through software that replaced the accountant you used to call. The work doesn't disappear. It just becomes invisible and unpaid. We call it convenience.

AI is accelerating this. Every task it makes "easy" for you is a task that used to be someone else's job, or didn't exist at all. The capability is real. But so is the hidden cost. And the question isn't whether things will eventually improve — it's what we lose in the meantime, and whether anyone's keeping track.

What happens to organisations?

There's a bigger question underneath all of this. When one person can do the work of five, what happens to how we've built our companies?

A developer building an app can now create the marketing video to promote it. Draft the legal contract to protect the IP. Register the copyright. When AI collapses every functional boundary, what happens to specialisation? What happens to culture when nobody's role looks the way it used to?

I don't have answers to those questions yet. But I think we need to be asking them now, before the shape of things gets decided without us.

Guardrails we choose for ourselves

About a month ago, I did something small that turned out to be one of the smartest things I've done recently. I built a custom skill in Claude (a way of customising how the AI behaves) that prevents it from writing first drafts for me. It forces me to think first, articulate my own ideas, before the AI gets involved.

I'd half-forgotten I'd done it. Then a couple of weeks ago I sat down with a paragraph of an idea and asked Claude to write the rest. It refused. It reminded me that I'd set things up to stop exactly that.

My past self had protected my future self from a lazy choice.

I think we need more of this. Not just for ourselves, but at every level. What would it look like if organisations built guardrails that protected their people from the pressure of limitless capability? What would change if we asked what we're actually trying to achieve with these tools, rather than defaulting to "more, faster" because we can?

The question is no longer can we build this? AI can build almost anything. The question — the only one that matters now — is should we?

It's getting dark. I'm going to stay in the woods a bit longer.