I Thought Attention Was the Scarce Resource. I Was Only Partly Right.

For the last year, I’ve argued that human attention is becoming one of the scarcest resources in the digital economy.

I still believe that. But I no longer think attention, by itself, tells us enough.

For years, digital platforms operated on a simple assumption: behavioral signals could tell us something about human intent. A view suggested exposure, a click suggested interest, and engagement suggested relevance. Entire industries learned to measure these signals and translate them into economic value.

That model worked better when synthetic activity was much harder to produce at scale. Synthetic activity itself is not new. AI changes its cost, sophistication and scale.

Machines can already generate content, comments, conversations, recommendations and increasingly complex digital behavior at enormous scale. As AI agents become more capable, they will also browse, compare, communicate, transact and execute tasks across digital systems.

That creates a deeper problem:

when a behavioral signal becomes cheap to manufacture, it becomes weaker evidence of human intent.

A thousand comments once suggested that many people had reacted. A completed digital journey suggested that someone had made a sequence of decisions. Those assumptions become much less reliable when machines can reproduce the visible signals of human activity cheaply and at scale.

This does not make attention worthless. It makes attention weaker evidence of intent.

A person can look without caring, click without intending to act, or spend time with something without making a decision. The more I think about this, the more I see value moving further downstream:

attention → intent → action → consequence

Each step can reveal more than the one before it.

Seeing a product is different from wanting it. Wanting it is different from trying it. And trying it is different from spending your own money, changing your behavior, learning something, recommending it to someone else or putting your reputation behind a decision.

This leads to an important distinction: not all actions carry the same amount of information.

The harder an action is to fake without real commitment, the more informative it becomes about genuine intent. Giving your time, spending your money, changing your behavior or accepting responsibility all require real commitment from the person making the choice.

The value is not in the action alone. It is in the intent, commitment and consequence behind it.

That is why I no longer think the next scarcity is simply attention. And I don’t think it is simply "human action" either. Human activity is not automatically economically meaningful.

What may become increasingly valuable is credible evidence of genuine human intent expressed through action.

This matters because AI is not only making content abundant. It is beginning to make activity itself abundant. And when activity can be manufactured, volume becomes a weaker signal.

Ten thousand interactions may tell us less than they once did. One real person spending their own money, changing their behavior or accepting responsibility for a decision may tell us much more.

The question therefore changes from how much activity happened? to something harder:

What happened, who did it, and what did it actually mean?

This goes far beyond advertising. Commerce depends on real decisions, education on actual learning, communities on genuine participation, governance on consent and accountability, and relationships on another person actually being there.

The more synthetic the environment becomes, the more valuable these distinctions may become.

Machines will perform many tasks better, faster and cheaper than humans. I don’t think the goal should be to preserve human involvement everywhere simply because humans used to do the work.

The more important question is where human participation still carries unique meaning, not because humans are automatically superior, but because some forms of value depend on identity, consent, accountability and consequence.

For years, we described the internet as an attention economy. That description made sense for the stage we were in.

But if AI makes both content and digital activity abundant, attention may no longer be enough to describe what becomes scarce.

Perhaps the next layer is closer to an action economy, one built around consequential human action rather than activity alone.

For years, the digital economy learned how to measure attention.

The next challenge may be learning how to distinguish activity from intent backed by real commitment.

Because when almost any digital signal can be generated, a much harder question remains:

What still tells us that a human being actually meant it?