The value of AI is not only what it can produce. The value of AI is whether it returns the human attention it consumes.
AI is usually evaluated by output.
Did it write the email?
Did it summarize the document?
Did it make the slide?
Did it answer the question?
That matters, but it is not enough.
The deeper question is:
What happened to human attention?
Did the AI absorb predictable work and return attention to the human?
Or did it create new unresolved surprise?
The Attention Cost Of Bad AI
Bad AI captures attention.
It does this in familiar ways:
Errors
Hallucinations
Vague drafts
Overconfident claims
Unclear provenance
Broken formatting
False completion
Repeated prompting
Review burden
Dashboard noise
Workflow interruptionBad AI may still produce something.
But if the human must monitor, correct, verify, reshape, distrust, and rework the output, then the system has created unresolved surprise.
It has not returned attention.
It has captured it.
The Attention Value Of Good AI
Good AI returns attention.
It absorbs predictable work and leaves behind completed artifacts.
It reduces the number of conscious remainders the human must resolve.
It does not merely move labor from one place to another. It lowers attention-demand.
A good AI system makes the next step clearer.
A good AI system reduces ambiguity.
A good AI system preserves context.
A good AI system finishes the boring parts without creating hidden review debt.
The human remains responsible for judgment, taste, intention, ethics, and direction.
But the system should not force the human to spend attention on what should have been absorbed.
The Formal Translation
In the attention model:
S(t) = Log(R(t))
C_W(t) = accumulated surprise
Attptr(t) = Nrm(C_W(t))
D_att(t) = |Attptr(t)|Bad AI increases unresolved surprise across the attention window.
Good AI reduces unresolved surprise across the attention window.
Bad AI raises D_att(t).
Good AI lowers unnecessary D_att(t).
That is the difference.
The New Product Question
The old product question was:
How much time does this save?
The better question is:
How much attention does this return?
Time saved is not enough if the saved time is filled with vigilance.
Automation is not enough if the human must watch the automation nervously.
Output is not enough if the output creates unresolved surprise.
The true value of a tool is measured by what it absorbs.
The Human Future
Human attention is the most valuable resource on the planet because it is the scarce conscious surface through which the future becomes history.
Everything that wants to shape the future wants access to attention.
Markets want it.
Platforms want it.
Politics wants it.
Entertainment wants it.
AI wants it.
But the best technology will not merely capture attention.
The best technology will return it.
That is the standard.
Bad AI captures attention.
Good AI returns attention.