Will artificial intelligence (AI) usher in an era of mass unemployment? For some of Silicon Valley’s most ardent tech evangelists, the answer is obvious. “There will come a point where no job is needed,” Elon Musk predicted in 2023. Two years later, Bill Gates took the discussion a step further, suggesting that within a decade AI would replace doctors and teachers, rendering people “redundant” in many places. It is not a new thesis.
Germany’s Der Spiegel, for instance, has returned to the story of technological automation pushing humans out of the labor market time and again — a theme it has featured on its cover at least four times since the 1960s, most recently just a few months ago, this time with artificial intelligence at center stage. Of course, as with every major technological shift, some jobs will disappear. But history suggests that work itself is unlikely to vanish. A few lessons are worth recalling.
1) New technologies automate tasks, not entire jobs
Technologies tend to mechanize or automate specific tasks, but they rarely eliminate the need for humans altogether. Economists call this complementarity: machines and people often end up working together, each doing what they do best.
That is likely to hold for AI as well. Consider education. AI-powered learning tools will not simply make teachers obsolete. More plausibly, they will free up time for what remains stubbornly difficult to automate: mentoring, motivating students, and managing the social dynamics of a classroom.
The jobs most at risk are those built around a single core task that a new technology can fully take over. In the past, that included coal shovelers on railways. Tomorrow, with autonomous driving, it may include taxi drivers. And yet, even for them, history suggests that new work will emerge, as the next lesson implies.
2) Technologies satisfy old wants more efficiently and create new ones
Human desires have long proven insatiable. When technology makes it possible to produce something better or faster, two forces tend to follow.
First, the product becomes cheaper and demand often rises. That, in turn, can increase demand for workers performing complementary tasks —tasks that machines still cannot do.
Second, as people gain time and disposable income, they develop new desires altogether. In the 18th century, few could have imagined that one day people would strap wooden boards to their feet for fun and pay to be pulled uphill by steel cables.
From a macroeconomic perspective, new technologies have rarely destroyed work outright. Instead, they have shifted it into new activities. It is no coincidence that six in ten jobs performed today did not even exist in 1940.
3) New technologies overwhelm people at first, but societies adapt
There are good reasons to believe the AI revolution will not bring about the end of work. Still, the fear is real and it should not be dismissed as mere technophobia.
Work, fortunately, is a source of meaning for many people. At its best, a job can feel like a calling. When occupations change or disappear, a social fabric unravels before it can be rewoven. From a distance, the transformation may look beneficial for society. For individuals, it is often a deeply personal burden.
Historically, that sense of dislocation has been manageable. Switzerland, in particular, has repeatedly managed to absorb technological upheaval without shutting down progress. Companies could respond quickly thanks to a flexible labor market, while social hardship was cushioned by a well-functioning safety net.
That institutional setup may help again with AI. But this time the challenge appears larger than before for one simple reason: speed, and scope.
AI does not merely affect one sector. It supports open-ended, cognitive work, touching many industries at once. And it spreads at breakneck pace because daily life is already deeply digital. ChatGPT needed just two months to reach 100 million users; the mobile phone, as a physical device, took more than 16 years.
The lessons of economic history may therefore be approaching a limit. The end of work? Elon Musk and Bill Gates are likely mistaken, but they may be right about the force and velocity of the disruption. And that makes adaptation, the third lesson, more challenging this time around.
This article was published (in German) in the “NZZ am Sonntag” on January 25, 2026.
