Something I wrote recently in another context and shared with some friends. Its inherent homage to clergyman Jonathan Swift will be obvious.

Have a good weekend.

Paul

A Modest Proposal for Preventing the Surplus Humans of the Post-AI Era from Being a Burden to Society, and for Making Them Beneficial to the Public

Being a Dispassionate Inquiry by Paul Kedrosky into the Efficient Disposition of Approximately 300 Million Displaced Workers, Conducted in the Spirit of Rational Optimism

It is melancholy to those who walk through the formerly great cities of this nation, or travel through the country, to see the streets crowded with persons formerly engaged in knowledge work. The streets, alleys, and shops are full of lawyers who once billed hundreds of dollars per hour to read codicils, radiologists who spent years learning to interpret the play of shadows on film, and, ironically, software engineers. 

These persons, having been assured repeatedly that automation would "augment rather than replace" them, and having acted on this assurance by acquiring advanced degrees, adjustable-rate mortgages, HELOCs, and 16-inch MacBooks, are now a public nuisance of considerable scale. They congregate in coffee shops where they share prompting tricks, like “five prompts to gain 25,000 followers in a week”, an enterprise roughly analogous to teaching a man to oil the guillotine.

I think all parties agree that this prodigious number of unemployed adults, and especially those who have also lost the life structure that work provided, is, in the present deplorable state of affairs, a very great problem. Whoever could find a fair, cheap, and easy method of making these people useful members of the economy would deserve so well of the public as to have their name inscribed alongside prior great benefactors of humankind, like the almighty Amodei.

But my intention is far from being confined to providing only for the knowledge workers. It is of a much greater extent, and shall take in the whole number of distressed humans whose cognitive labor—the one faculty that distinguished them from prior automation targets—has been priced at approximately zero by offshore systems running on state-subsidized electricity.

I shall now therefore humbly propose my own thoughts, which I hope will not be liable to the least objection.

What I propose is this: that the 300+ million shelled-out Americans whose jobs have been automated or substantially degraded be productized. 

Specifically, I propose the following three programs: 

First, that surplus humans of an emotive disposition be redirected toward witnessed suffering services, in which human grief, confusion, and economic anxiety are performed for the benefit of AI training pipelines that require emotionally authentic data at scale. The subjects would thus monetize their distress, which would improve the models’ ability to simulate extreme emotions. Value is created. I see no ethical difficulty that a well-written terms-of-service cannot resolve.

Second, that we establish a National Register of Overly Humans, a reserve population of persons who are, by training or constitution, incapable of being well-simulated. These are people with the specific combination of inconsistency, irrationality, and unearned confidence that defines genuine human judgment. These persons, currently a liability in any well-AI-ed society, would be removed to reserves where they could be studied as heritage assets, but not contaminate the broader populace.

Third, and most ambitiously, I propose we consider revisiting converting shelled-out humans to base assets. While some years ago paperclips had been suggested, with paper no longer in wide use, this seems wasteful. I suggest considering copper and iron, given their abundance in the human body and their role in Holy Data Centers. A population of 350 million humans could be converted, assuming minimal losses, to 30 tonnes of copper and 1 million tons of iron. The latter amount is enough to build 25 data centers, which would be a boon, far more societally useful than mere prompt-memeing in coffee shops, or wandering the streets, slack-eyed and meme-seeking. Granted, it is a one-time use, but we must think of the discounted present value of such people’s costs to the future, and weigh that against the myriad benefits of more sanctified Data Centers in the present. Again, I see no ethical difficulty that a well-written terms-of-service cannot resolve.

I can think of no consequential objection that will be raised against this proposal, unless it should be urged that the number of humans vastly exceeds market demand for witnessed suffering or heritage museums, or that we need no more Data Centers. To this, I can only say: I am aware of no principle in economics that requires supply to find demand. The market for buggy whips also contracted, and we do not speak of it. 

Some persons of a desponding spirit are in great concern about the vast number of displaced workers who are aged, diseased, or otherwise non-productizable. I am not in the least pain upon that matter, because it is very well known that they are every day dying and rotting, by cold and famine, and filth and vermin, as fast as can be reasonably expected. And as to the younger laborers, they are now in almost as hopeful a condition: they have sold their vehicles to the fleet operators, their apartments to the short-term rental aggregators, and their attention to platforms that have since replaced them with agents that do not require engagement incentives and do not experience loneliness.

I profess that I have not the least personal interest in endeavoring to promote this necessary work, having no other motive than the public good of my country, the advancement of our almighty AI industry, the relief of the quadrillions in shareholder value created by the transition, and the speedy disposition of displaced persons who have no other prospect of ever otherwise finding themselves in the revenue model.

—Paul Kedrosky, ITYOOA 16