Next time you're stuck on a question, ask nothing for ten minutes. Let it churn. You'll likely notice two things: that it's uncomfortable, and that something is happening. This uneasy state has a name, and we're quietly removing it from our lives without a second thought.
We've written a lot about what AI saves us. Let's talk about something it's making disappear, something that shows up in no balance sheet because no one ever counted it as a resource.
What boredom used to produce
Boredom has a bad reputation. It's described as a void, wasted time, something to fill. Research in cognitive psychology suggests the opposite: it's a distinct state of functioning, and it's productive.
When the mind isn't occupied by an external task, it doesn't stop. It starts to wander, to connect seemingly unrelated things, to circle back to unresolved problems. It's in this state that most unexpected associations occur—the ones we say came to us in the shower or on a walk.
Boredom also produces something subtler: a tolerance for uncertainty. Sitting with an unanswered question, accepting not knowing for a while, is a skill that needs upkeep. It serves well beyond intellectual matters.
What AI changes, precisely
Smartphones had already largely eliminated boredom, filling every gap with content. AI does something different, and deeper.
It doesn't just fill dead time—it fills uncertainty. Before, an unanswered question stayed unanswered for a while, for lack of an available interlocutor or an easy source. That delay was when you refined the question, searched on your own, sometimes found the answer.
That delay can now drop to zero. And it drops to zero by default, without any decision on your part: asking has become easier than thinking.
This ties into what we examined in our article on cognitive delegation, with the important caveat that this work is correlational. The most interesting finding, recall, was about order: those who had first worked without assistance retained better abilities. Boredom might well be the name for that preliminary phase we now skip.
What not to conclude
It would be easy and dishonest to sing a nostalgic praise of boredom. Two nuances matter.
Not all boredom is fertile. Waiting in an administrative queue produces nothing. Creative boredom requires an available mind and a latent problem, not just empty time. Much of what AI removes, no one will miss.
This isn't new. Every comfort technology has removed a form of friction whose loss some lamented. Printing removed the effort of memorisation, the calculator that of mental arithmetic. These losses freed attention for other things. There's no principled reason this one should differ.
The question, then, isn't whether boredom should be preserved for its own sake, but whether what it produced has a substitute.
What makes this case particular
One difference deserves to be stated. The calculator removed a mechanical operation. AI potentially removes the step where you formulate the problem.
Yet we argued in our article on the art of asking a good question that this step has become the core skill. So there's a tension: the tool that makes the question decisive is also the one that spares you from working on it.
This isn't inevitable. It's a trade-off, and it plays out with every use.
What to take away
There's no rule to hand out, and I'd be wary of any article telling you how many minutes of boredom to preserve per day.
What seems right is to distinguish two situations. When you're looking for information, ask without hesitation: no one gains from not knowing a date. When you're trying to understand something, or find an idea, waiting has value. It's unpleasant precisely because it's working.
There may be one simpler reason left to protect these moments. Boredom is one of the few states where you expect nothing from anyone, not even from a machine. It's a small thing, and it's becoming rare.