Few economists have left behind sentences that endure for decades. One of them comes from Robert Solow. The American Nobel laureate remarked in the late 1980s: "You can see the computer age everywhere but in the productivity statistics." It is a quote that sticks, and it often comes to mind amid the general excitement over artificial intelligence (AI).
Most people are currently debating whether AI will one day wipe out humanity, or whether part of the hype mainly serves the share prices of the companies involved. But another question deserves attention, precisely because it touches everyday life: Is AI already raising productivity, meaning the value created per hour of work? Or is the new era, for now, showing up more on the front pages than in the statistics?
Faster is not necessarily more productive
At first glance, many would say the answer is obvious. How often has AI taken over a tedious task, whether creating a slide deck, drafting an email or gathering information? Studies do show that individual tasks can be completed significantly faster with AI. The decisive question, however, is whether this also makes companies as a whole, and ultimately the entire economy, more productive. Here, the evidence is less clear.
Research on the subject is still in its early stages, and much of it still relies on surveys and case studies. Early firm-level studies, though, suggest that perceived productivity gains may be larger than what has actually been measured so far. And an analysis by researchers at the U.S. Federal Reserve showed this summer that, at the macroeconomic level, investment in AI is clearly visible, while broad productivity gains are still far less so.
Why, as with computers, is the productivity boost taking so long to materialize? A possible answer emerges when one takes a closer look at today's office life. There are several signs that completing an individual task more efficiently does not yet make the overall system more productive.
Implementation is a cultural task
New slide decks and reports can now be produced in minutes. But someone still has to read, assess and, if necessary, improve them. What saves time for the creator creates new work for the recipient. If the recipient in turn uses AI to evaluate the output, the cycle can spread from department to department. The result is often frantic activity everywhere, but by no means an increase in the organization's productivity.
Completing a single task efficiently does not automatically make the system more productive. That may explain why AI is visible everywhere, yet its effect on the productivity of the economy as a whole is far less apparent. It is a pattern reminiscent of Robert Solow's famous observation. This may be a sobering finding. But as with computers, a productivity surge will probably arrive in the end. The question is how it could be accelerated already today.
For now, it is worth reminding ourselves that busy activity is no sign of productivity at all. It can even point to the opposite. The focus should therefore shift away from individual tasks and toward the content of work and the structures around it. Historically grown processes must be questioned and, where necessary, redesigned from the ground up. Implementing AI tools in an organization is therefore not simply an IT project. It is a strategic and, above all, a cultural task.
Without a plan, the costs rise
This sounds like a challenge for top management, and it is, but not only that. In the end, everyone is called upon, in their own daily work. AI radically reduces the cost of countless tasks. Interestingly, this same dynamic raises the cost of lacking priorities. Those who work without a plan waste not only their own resources and computing power, but increasingly those of their colleagues as well.
Finally, let us recall management guru Peter Drucker, who emphasized the point roughly a quarter century before Solow's productivity observation: There is nothing so useless as doing efficiently that which should not be done at all. With technological progress, this insight has only grown in importance. Whoever uses AI to efficiently complete many wrong things is anything but more productive. They have simply built a treadmill that spins at enormous speed.
This article first appeared in the newspaper "NZZ am Sonntag" on October 4, 2026.