If we set aside the promotional optimism of technology companies and apocalyptic fantasies, I see the main risk of the near future not in the idea that “AI will destroy humanity,” but in the fact that it may very quickly change the economic value of the average person.

In its time, the Industrial Revolution replaced human muscle with machines, but it left enormous demand for the human mind: accountants, engineers, managers, designers, operators, programmers, lawyers, journalists. The current revolution is, for the first time, moving into precisely this territory on a massive scale. At the same time, robotics is beginning to take over physical labor as well. In other words, the two waves may converge.

I think the consequences may look roughly like this.

First, a significant share not of professions themselves, but of jobs within professions, will disappear. This will not necessarily lead to the complete disappearance of accountants, for example. More likely, one accountant using AI will do the work of five. One designer will do the work of three. One programmer will do the work of what used to be a team of several people. One lawyer will process hundreds of documents.

And this may be even more dangerous than the complete disappearance of a profession, because outwardly the economy will look much as it always has: companies will operate, products will be created, GDP may even grow. But far fewer people will be needed to make all this happen.

Second, the middle class will suffer the most. Twentieth-century automation hit factory labor hard. AI is primarily hitting office work. And office work has been the foundation of the urban middle class for the past several decades.

Copywriters, translators, junior programmers, analysts, accountants, support operators, recruiters, designers, paralegals, lower-level managers — this is an enormous segment of the population.

And I consider the situation especially dangerous for young people. Companies may stop hiring junior specialists because AI can perform their work. But then, ten years later, a paradox arises: where will senior specialists come from if no one has gone through the junior stage?

Third, productivity may increase fantastically while the well-being of the majority does not. This is a fundamental point.

Imagine that, thanks to AI, a company produces the same product with 200 employees instead of 1,000. Economically, this is excellent. But the question is who receives the gains. If they go to the owners of the company, we will end up with extraordinarily wealthy owners of capital and a mass of people who are increasingly unnecessary to the economy. If part of the gains is redistributed through wages, taxes, a shorter working week, social dividends, and so on, then the technology may genuinely make society richer.

In other words, AI by itself creates neither a utopia nor a dystopia. It sharply increases productivity. The question of ownership then determines who receives the benefits.

That is precisely why I believe the central conflict of the twenty-first century may not be between humans and machines, but between the owners of automated capital and people who sell their labor.

Fourth, labor may cease to be the primary mechanism for distributing income. Today, the logic of society is simple: if you want to consume, earn money; if you want to earn money, sell your labor. But if the economy simply does not require that much human labor, this formula begins to break down.

You cannot endlessly demand that people retrain. If trucks are driven by robots, documents are drafted by AI, stores are automated, warehouses are automated, call centers are automated, and software code is generated to a large extent by machines, there is no mathematical law requiring the same number of new human professions to emerge. They may emerge. But they may also fail to emerge in sufficient numbers.

At that point, various forms of guaranteed income, negative income tax, public services, or a “social dividend” will cease to be a left-wing utopia. They may become simply a technical necessity for sustaining a consumer economy.

Because a very simple problem arises:

robots produce goods — but robots do not buy them.

Fifth, inequality may become much greater than it is today. The digital economy already operates according to a winner-takes-most principle. A single good software product can serve a billion people. AI intensifies this effect.

A company of the future may potentially be worth hundreds of billions of dollars while employing a relatively small number of people. This means that the link between the scale of economic activity and the number of people who receive wages from it will weaken.

A strange society of enormous material abundance may emerge in which a significant part of the population feels economically redundant. And this is no longer merely an economic problem.

Sixth, there will be a crisis of meaning. This is discussed far less than unemployment.

For a modern person, work is not only about money. It provides status, structure to the day, a social circle, and a sense of competence — “I know how to do something,” “someone needs me.” You can pay people enough money to live a normal life and still end up with a generation experiencing an acute sense of uselessness.

That is why I am not convinced that a society of mass leisure will automatically be a happy one. For a small proportion of people, freedom from work will mean books, science, art, travel, and family. For others, it may mean endless videos, games, alcohol, virtual worlds, and primitive stimuli.

There may even emerge a kind of class of economically unnecessary people, whom the system does not oppress in the old sense — it simply needs almost nothing from them. I consider this psychologically very dangerous.

Seventh, the power of those who control AI infrastructure will increase dramatically. In the past, land conferred enormous power. Then factories did. Then oil and financial capital. In the twenty-first century, that power may come from computing capacity, models, data, energy infrastructure, and robotic production. Whoever owns a system capable of replacing the labor of millions of people gains not merely a large business — they gain structural power over society. And this opens the door to very different political models, ranging from entirely liberal ones to extremely authoritarian ones.

But there is another side to this. I can entirely imagine that, 30–50 years from now, our descendants will look at the five-day, eight-hour working week the way we look at six-day factory labor in the nineteenth century. It may seem absurd that a person spent most of their conscious life filling in Excel spreadsheets, drafting standard contracts, driving a truck, or moving goods around a warehouse simply in order to earn the right to housing and food.

If machine productivity becomes high enough, freeing human beings from necessary labor may in itself become one of the greatest achievements of civilization. The only problem then will be the transition.

And it is precisely the transition that, in my view, will be the most painful. Technology may change within 10–15 years, whereas education, legislation, culture, social institutions, and people’s ideas about their own value change much more slowly.

So my baseline scenario is as follows. I do not believe that everyone will lose their jobs. A more realistic outcome is that a relatively small percentage of highly productive people, together with AI and robots, will generate an ever larger share of economic value, while society spends several decades painfully trying to decide how to distribute the wealth that has been created and what to do with millions of people whose labor is no longer economically necessary.

And if this question is resolved well, the result may be the richest and freest society in history. If it is resolved badly, we may end up with a very wealthy society containing an extraordinarily large number of poor or dependent people.

Knowing human nature and history, I consider the second outcome more likely.