The original note, as written at the time
In the peak of the AI era, it may be possible for each one of us to run a successful business because our AI agents will be helping us to attend to the needs of others perfectly. We shall then dedicate ourselves to some kind of humanitarian or civil duties.
Ongoing development
At the peak of the AI era it may be possible for each of us to run a successful business, because our agents will attend to the needs of others so well that we can dedicate ourselves to civil and humanitarian duties.
I want to take that seriously rather than treat it as a fantasy, because the mechanism is plausible and the consequences are not what people expect.
If the operational burden of a business — correspondence, scheduling, accounts, follow-up, the paperwork of being trustworthy — is carried by an agent, then the barrier to running something is no longer capital or time. It is whether you know something worth doing for somebody.
That is a genuine liberation, and it would produce an enormous number of small, competent, unremarkable businesses: one person, an agent, and a trade.
But the argument only holds together because of the second clause. The point is not that everybody becomes an entrepreneur. The point is that the time released is meant to go somewhere, and the place it is meant to go is public life — the school committee, the clinic, the water project, the things that currently go undone because nobody with competence has an hour to spare.
Without that clause it is just a labour market with more precarious units in it. With it, it is a claim about what a society would do with a sudden surplus of capable attention.
That is testable, and it should be tested honestly. If people free of operational work mostly buy more leisure, the theory is wrong and we should say so. If a meaningful number turn to the unpaid work that has been starving for decades, we would have found something better than productivity.
The agent is not the interesting part. What we do with the reclaimed hours is.