AI is framed as a productivity tool. Write faster, design faster, build faster. The promise is straightforward: the same work, done in less time.

And at first, it feels true. You generate something in seconds that would have taken hours. You move past blank states instantly and get to something usable without much effort. The early part of the process compresses in a way that feels almost unnatural.

But that compression doesn’t reduce the work. It changes where the work lives.

What actually happens is harder to notice, because the gain in speed creates a new kind of behavior.

The moment something becomes easy to do, you start doing more of it.

You don’t just write one version, you try multiple. You don’t just land on a direction, you explore variations, refinements, and edge cases you wouldn’t have considered before.

This isn’t inefficiency. It’s a natural response to reduced cost.

When iteration becomes cheap, exploration expands to fill the space. The system doesn’t encourage completion, it encourages continuation.

Before AI, friction acted as a boundary.

You stopped refining because it took time. You shipped because another iteration wasn’t worth the effort. You accepted “good enough” because better had a real cost attached to it.

Those constraints weren’t just limitations, they were decision-making tools. They created a natural end to the work.

With AI, those boundaries disappear.

Another version is always available. Another improvement is always one prompt away. The cost of iteration approaches zero, and with it, the reason to stop begins to fade.

At the same time, the bottleneck doesn’t disappear. It moves.

When execution becomes instant, judgment becomes the limiting factor. You’re the one deciding what’s better, what direction to take, and when something is finished. The system can generate endlessly, but it has no concept of sufficiency.

Speed increases output. Output increases options. Options increase decisions. And decisions don’t scale the way execution does.

So even though each step is faster, the total work isn’t necessarily smaller. In many cases, it grows because the number of possible paths expands faster than your ability to evaluate them.

AI didn’t remove work. It removed the constraints that made work finished.

Less time is spent creating. More time is spent evaluating, selecting, and deciding. The effort shifts away from execution and toward judgment, where the cost hasn’t meaningfully decreased.

That shift is easy to miss because output is so visible, while decision-making is not.

From the outside, it looks like productivity has increased. There’s more output, more ideas, more visible progress.

But from the inside, the experience is different. The work doesn’t feel lighter. It feels continuous, with fewer natural endpoints and more pressure to keep going.

When friction disappears, so do the boundaries that made work end.