It Depends on Us
Publish date 26-07-2026
Despite my efforts to use Artificial Intelligence (AI) consciously to improve my work processes and work-life balance, I have the feeling that I am working more and more. Falling victim to the illusion of completing the multiple and diverse tasks that characterize my job in less time, the things to do end up increasing instead of decreasing. But wasn't the promise of Artificial Intelligence to increase productivity and, finally, to work less? Caught in the embarrassment of not being able to explain this paradox to myself, I sought comfort, as I often do, in economic research.
A study published in the Quarterly Journal of Economics, one of the most prestigious economics journals, analyzed the introduction of generative Artificial Intelligence tools in a large call center: a very specific context, in any case, and quite different from that of my profession (university professor). The results are clear: AI increases productivity. On average, performance improves by about 14%. But the most interesting data is another: the benefits are not distributed evenly. Less experienced workers improve much more than the best ones, narrowing the performance gap.
Good news, I would say. In this context, Artificial Intelligence does not replace the worker, but functions as a true "digital tutor." It suggests answers, incorporates best practices, and makes accessible skills that were previously unequally distributed. Rather than simply accelerating work, it modifies its structure and processes, making certain activities more standardized and replicable.
Based on these results, the most immediate intuition is that higher productivity can translate into fewer hours worked. If more is produced in the same amount of time, one could work less for the same output. At least that is what I learned when studying economics. However, the observed reality is more ambiguous. Recent studies based on data from the American Time Use Survey show that, in sectors most exposed to AI, after the introduction of tools like ChatGPT, people tend to work more, not less. On average, there is an increase of about three hours per week, accompanied by a reduction in free time.
Thus, the paradox that torments me resurfaces: if technology makes work more efficient, why does it not free up time? The answer, from what I gather, is that efficiency does not automatically determine how the saved time is used. That time can be reabsorbed in the form of new activities, more ambitious goals, or higher standards. In some cases, Artificial Intelligence does not replace work, but expands it or intensifies its pace.
On one hand, it is the socio-economic context that pushes us toward the idea that "more is better"; on the other hand, our own tendency to seek satisfaction and recognition – often on that never-stopping treadmill of ambition – leads us to always do more. In this context, the increase in productivity facilitated by AI risks working against free time, reinforcing the feeling of being able to – and in part having to – further increase the workload.
The picture that emerges is therefore less simple than the dominant narrative. AI increases productivity, but does not automatically lead to a reduction in work. Its effect on personal well-being is ambiguous: it can make work lighter or more intense, more accessible or more controlled, fairer or more demanding.
It is useless to blame AI: the ultimate result depends on how we choose to use it. The reduction in hours worked is not an inevitable consequence of technological progress, but the result of organizational, social, and personal choices. AI broadens possibilities; whether these will translate into more free time or greater productive pressure will depend on how the benefits of this transformation are distributed, as well as on the meaning we attribute to work and how it is integrated into people's lives.
Pierluigi Conzo
NP April 2026




