Attention in the Age of AI. When Everything Is Abundant, What Remains Scarce?

I used to pay for a training app because building one myself would have been absurd. Then I found an open exercise dataset, gave an AI agent the rough shape of what I wanted, and spent an afternoon turning it into something good enough for my own use.

The result was not a startup and it did not need to be. It was a small private tool that solved a small private problem. What stayed with me was how quickly the constraint had moved. I did not need to ask whether I could build it. I had to ask whether it deserved an afternoon of my life.

AI is making code, text, images, analysis, and prototypes abundant. Production has not become free, but it has become cheap enough that many ideas can acquire a visible form before we have decided whether they are worth pursuing.

That changes the job.

The bottleneck moved upstream

At a product conference, Claire Vo argued that code is abundant while markets are not. Her point was practical. A team can now build much more than it can sell, support, or persuade anyone to use. A large codebase no longer proves that a company has found a real problem.

The same shift appears outside software. An AI can produce ten strategies, thirty headlines, and a complete report. The scarce part is deciding which problem deserves the report, which claim can survive contact with reality, and which of the ten strategies should consume another month.

Attention here does not mean concentration in the self-help sense. It is the act of assigning limited human time to one possibility and leaving thousands of others untouched.

AI can widen the field of possibilities. It cannot make that allocation harmless. Every yes still spends a piece of a finite life.

Verticality is a shortcut for scarce attention

Recruiters prefer candidates who have already worked in the same field. Platforms prefer creators who keep publishing in the same category. Investors prefer companies they can place inside a familiar market. These systems are often described as rewarding depth.

Sometimes they do. They also use verticality as a cheap filter.

A recruiter cannot reconstruct every candidate's ability from first principles. A platform cannot carefully read every new account. Continuity becomes a proxy for quality because attention is too scarce to inspect everything directly.

This creates an awkward tension. Selection systems reward a clean vertical line. New value often comes from crossing lines that were previously separate.

My own work has moved across climate policy, international cooperation, product design, AI systems, and public writing. On paper, that can look unfocused. In practice, the crossings are where many of the useful questions appear. What does an AI product owe the person whose work it automates? How should a system preserve judgment when efficiency becomes easy? Those questions do not belong neatly to one job category.

The difficulty is that a cross-disciplinary advantage is expensive to evaluate. A vertical label is cheap.

Earning attention is different from capturing it

When production becomes abundant, it is tempting to conclude that distribution is everything. That quickly turns attention into a game of hooks, frequency, and reach.

Capturing attention can make someone look at a thing. Earning attention gives them a reason to return.

The distinction matters because AI can industrialize the first one. It can generate more posts, test more openings, and fill every channel with competent material. The result may be more visibility and less trust.

Earning attention still depends on repeated judgment. Did this person choose a problem that matters? Did the work contain something that could not be produced by expanding a prompt? Did the writer respect my time?

That is why restraint is becoming a productive skill. Choosing not to publish the sixth version, not to build the extra feature, and not to send the unnecessary memo can create more value than another round of generation.

AI gives more people the power to make things. It also removes the old excuse that making the thing was the hard part.

The harder question is now visible. What deserves our attention, and what have we done to deserve anyone else's?

Source note on Claire Vo's talk, The New PM.