The effects of artificial intelligence on work are the favorite topic of those podcasters who interview prophets of the apocalypse. There exists a roster of guests, always the same ones, often defectors from big tech (Google, Microsoft) who have reinvented themselves as sorcerers, “doomsters,” or gurus: they want to warn us about the risks of AI, and their specialty is making alarmist long-term predictions. The discourse oscillates between dystopia – unemployment, poverty, rule by machines – and utopia – universal basic income, machines working in our place, free access to resources. This polarized discussion, largely devoid of data, also appears in traditional media. It’s a technological revolution whose effects we don’t yet see; in reality, no one knows anything for certain. “It’s important that research creates data through experiments, to give substance to this debate, to go beyond the chatter,” says Emilio Calvano, professor at Luiss in Rome and fellow at the Einaudi Institute. He specializes in industrial organization and competition in the technology sector. He authored research that as early as 2020 demonstrated that the artificial intelligences of competing companies learn to collude, agreeing on prices to maximize profits at the expense of consumers. “There’s still little scientific literature, but there’s a great need to quantify things, even those we have a correct intuitive perception of.”
Calvano gives me what he considers the essential coordinates for understanding the effects of this technological innovation on work, which is only beginning – very few companies have integrated it into their production processes so far, not only in Europe but also in the United States. Previous industrial revolutions brought automation. “Man told the machine: if these conditions exist, then act this way.” With artificial intelligence, instead, we give machines a goal, a what not a how. This is machine learning: “Knowledge is the output, not the input. This exposes a much broader area of tasks that previously was thought impossible to delegate to machines and these tools.” What does “expose” mean? “In labor economics, we’ve started to consider work—that is, occupations—as a set of tasks. The idea is that within these occupations, complete automation of some tasks will occur.” But in others, humans will work using AI. “Occupations will restructure themselves in a way that allows workers to dedicate more time to tasks in which they have a comparative advantage over machines.”
The tasks replaceable by machines are those that are “exposed.” Calvano gives a concrete example. “A machine shown many X-rays can identify patterns invisible to the human eye. The radiologist identifies based on their knowledge, and the machine highlights elements that support the diagnostic process; the combination generates a superior output. Productivity increases and therefore salary (of the doctor) increases.” Supply can be increased. “In this scenario, a specialist visit costs less; for equal costs, healthcare can be offered to more people.” We glimpse for a moment the utopia of medical care for all, goodbye waiting lists. “It’s a textbook way of thinking about the problem. Productivity increases, which means salary increases, access to the profession increases (because higher salaries attract more doctors), this drives down costs, more demand is generated for these services, and so on.” We’re far from the apocalyptic scenario where machines take our place and we disappear. On the connection between greater productivity equals higher wages: “You tell one worker to dig a hole with a shovel, another to dig a hole with an excavator. The first takes two days, the second takes an hour. For equal hours worked, who has the higher salary between these two?” In other words, greater productivity raises wages, if there isn’t an infinite line of workers. The doubt remains that billionaires will emerge from this process with infinite bargaining power over wages, thus ending up appropriating most of these productivity gains.
We know that some jobs will simply disappear – perhaps because we’ve listened to too many apocalyptic podcasts. It’s called the displacement or substitution effect. “There are more exposed jobs, but it’s not the first time a technological innovation has somewhat restructured the world of work.” The point is to understand in which direction the restructuring is going. At the NBER conference at Stanford, work was presented demonstrating that 75.2 percent of at-risk roles in the American workforce are occupied by women. The data reflects their concentration in office tasks. So AI will penalize the female employment rate and the wage gap. The positive news is that most at-risk workers have “above-average adaptive capacity.” But having to change work sectors is no small matter; female workers on average have more volatile careers, more interrupted ones (perhaps also by maternity), and lower salaries.
Will there be a great economic boom, as after all innovations, growth and abundance of new jobs? Not quite. According to Daron Acemoglu, Nobel Prize winner in economics in 2024, AI’s impact on growth and productivity over the next ten years will be rather limited. Calvano shows a graph of productivity effects on a call center after AI adoption (measured by number of problems solved): it improves by 15 percent, with a more pronounced effect on less experienced workers. Not quite a scenario where we all go to the beach with universal basic income (the utopian scenario of podcast gurus). But it’s really too early to say. For the implementation of these technologies, we must wait. New companies will be born, created around AI, that will enable its adoption. According to Calvano, Europe’s delay on this front isn’t so marked: the development of applications—that is, the part of the supply chain that uses models (ChatGPT, DeepSeek) and packages them to create tools suited to particular purposes – is still at an initial level; the game is still open.
He concludes with a thought on work. According to him, it’s wrong to think only in terms of suppression of old occupations and creation of new ones; complementarity will be fundamental: “I teach at university; I don’t think my students will be left without work. I think the world will be divided between those who know how to use these machines and those who don’t. To perform any job at your best, you’ll need to know how to use these tools.”
Emilio Calvano is an economist, a professor at Luiss Guido Carli University, and a fellow of the Einaudi Institute.
