Will the AI Revolution Foment a Revolution?

W hen people debate the potential economic and social impact of AI, there are two basic perspectives.

In the more optimistic view, AI will ultimately be like other technological advancements throughout history. It will cause disruption that may be felt hardest by certain professions, but it will gradually be integrated into our lives as a tool to make us more productive and better at our jobs, and it will free up time for leisure. If this plays out, a few decades from now, we’ll be able to objectively weigh the positives and negatives of AI just as we can with the internet, and the march of history will proceed more or less as it has.

But there is also a doomer view. Under such a scenario, AI advancements won’t proceed as gradually as past technological progress, a pace that gave society time to adjust. Instead, there will be vast leaps. Within a few years, jobs that have to this point required significant knowledge and experience will be done faster and cheaper by a bot. This will rapidly make large percentages of the white-collar workforce unemployable.

On Tuesday, AI exec Matt Shumer posted a viral essay making a version of this argument. He compared the current time to February 2020, when some people who were following things closely could tell that something major was up with Covid, but most people didn’t realize it. He explained that those who are working in the tech industry are more aware of it than anybody because they are the first to be replaced. “I am no longer needed for the actual technical work of my job,” Shumer wrote. “I describe what I want built, in plain English, and it just . . . appears. Not a rough draft I need to fix. The finished thing.”

I am not going to predict which scenario is more likely. I don’t have the technological expertise to do so, and even those who have a much better understanding of AI than I do disagree amongst themselves.

But I do have a certain understanding of history and of contemporary U.S. politics. And what I worry about is that most thinkers, especially on my side, are betting on the first scenario without preparing for the second.

To state it more plainly, if we’re in a situation in which there is sudden and massive unemployment among white-collar workers, there is no obvious answer that those of us who believe in the free market and limited government have to offer.

I had first wanted to write on this subject last fall, but I thought it would be much stronger if I could present a range of free market responses to such disruption. As somebody who spent a decade advocating free market ideas in the complicated area of health-care policy, however, I’ve been stuck because I did not come across a plausible answer.

Most of the literature I have come across about AI from a free market perspective is more focused on making the first case I presented — the one cautioning against doomerism and arguing, essentially, that every time there is technological innovation, people warn about the coming calamity, but we have always ended up better off in the long run.

I have no doubt that this is a perfectly plausible possibility — and maybe the most likely one. But the question that I can’t get past is: What if the other scenario plays out?

It’s obvious what the answer will be from the other side. Democrats and big-government Republicans will say we need universal basic income — or some even worse form of government intervention. Yet it isn’t clear what free market folks could say.

What would be unique about the AI doomer scenario is that it would disproportionately hit higher-educated, white-collar workers, whereas past technological innovations have tended to hit blue-collar workers more. I say this not to be dismissive of the past struggles of displaced workers or to imply that white-collar workers’ losing their jobs is more important. I say this to underscore that such a disruption would be substantively unique in two major ways.

First, it’s much more difficult to replace lots of high-salary jobs that are quickly eliminated by technology. Let’s say, for instance, you were an ice deliveryman who became unemployable after the mass adoption of refrigerators. The transition to home refrigeration happened over decades, so you could eventually learn how to service refrigerators or perform other manual tasks to replace your income. It’s much harder to do that for somebody who built up expertise in a white-collar profession that pays $150,000 and is suddenly no longer necessary because of AI. Or, if you’re a law school graduate from a mid-tier school carrying hundreds of thousands in debt who can’t get a job because firms are using AI to perform document review or other tasks traditionally done by more junior attorneys. To be clear, I’m not saying that all lawyers would be replaced by AI. But what if, for certain tasks that traditionally required 20 lawyers, a firm only needs five lawyers who feed prompts to a large language model? And what if every law firm is downsizing around the same time, flooding the market with out-of-work attorneys?

The second issue I would raise is that a lot of famous revolutions have been fomented by relatively elite classes rather than the hoi polloi. This was certainly true of the American Revolution, which was led by merchants, landowners, and lawyers. In the French, Russian, and Chinese communist revolutions, there was certainly unrest among the lower economic classes. But relative to the general population, the leaders of those revolutions were well-off or products of elite education. Even Mao, who was a peasant, came from a family that was comparatively prosperous for somebody of that social class — and while he didn’t receive a formal education at a prestigious foreign university, he did receive more education than was typical at the time and was well-read. Plus, early communist leadership included many urban intellectuals.

A massive technological change in which the upper-middle class is the hardest hit is something that we haven’t quite experienced in this country. As things stand, we are at a fragile time in the history of our republic. There are large contingents within the bases of both parties that have become frustrated with the inability of their elected officials to advance their agenda when in power, given the roadblocks created by Congress (especially the Senate) and the courts. Depending on who is in power, on both the hard right and hard left these days, you don’t have to search too far to find people arguing that the president should ignore Congress, ignore the courts, and implement his agenda. This is during a time of relatively low unemployment and prosperity. Throw into the mix a historic technological upheaval that is laser-targeted at an educated and well-connected class, and I fear that if the apocalyptic scenario plays out, it will be more destabilizing to our politics than anything we have previously experienced.

We recently saw with the election of New York City Mayor Zohran Mamdani how downwardly mobile individuals can embrace radical ideas that, even in a left-wing city, would have been unthinkable a few years ago. It’s this class that is most prone to economic resentment, because they are more likely to have grown up or gone to school with the super-rich classes and see themselves as just as capable.

Implementing UBI or some other socialist scheme to ameliorate the effects of an AI revolution might be getting off easy. We could be looking at blowing up the filibuster or eliminating the Senate altogether, ignoring court rulings, or packing the Supreme Court. In the most extreme case, in which elite anger combines with popular violence, we could end up with a revolutionary fervor that sweeps through the nation and topples the republic.

Again, I am not saying that this is the most likely scenario. On most days, I am more convinced by the arguments in favor of the more optimistic scenario in which AI gets integrated into our lives like past technological improvements. But I do think those of us who believe in limited government should start thinking about what sort of answers we would provide if the worst-case scenario plays out.