Three Months Writing about AI and Work: In the End, It Resulted in a Book
This article is different from the usual. Since February, IA Upgrade has been a weekly appointment, usually focused on a current news item or data, told with a skeptical eye toward industry hype and with an evident discomfort for press releases that present themselves as analyses. Today, however, I am writing about something I have done in parallel with the column, and I feel it is important to share it transparently.
A methodological premise. When writing weekly about a rapidly evolving topic, the only sustainable strategy is to read a great deal. You end up accumulating academic papers, institutional briefings, research reports, statements from CEOs, legislative traces, industry data. A portion of this material goes into the column. At some point, I realized that the material left out of the pieces had become more substantial than the material used, and I gathered all these insights into an archive that, as the days went by, became enormous, discovering that many of the themes addressed in separate articles were actually chapters of the same reasoning.
In short, in the end, I wrote a book. It’s called Humans for Now -- Why This Time Work Isn’t Coming Back, and it is available on Amazon in both paperback and Kindle versions. As I mentioned, it largely emerged from the documentation work I do for these pages. The advantage of the book format is that the topics interact with each other, taking on a different scope and consistency.
The book starts from an observation: the established script of technological revolutions over the last two centuries tells a fairly simple story. The machine replaces a job, the production system creates others, usually more qualified and better paid. Weavers became factory workers, factory workers became white-collar workers, and white collars morphed into knowledge workers. The mainstream optimistic narrative uses exactly this historical trajectory to argue that it will be the same this time; you just need to wait and re-skill. The point is that previous technological revolutions automated physical labor, repetitive actions, calculations, and archiving, while the current generation of AI systems is automating the very cognitive skills that previous replacement trajectories had built as a refuge zone. Reasoning, writing, analyzing, synthesizing, re-synthesizing. This changes the meaning of the statement "work always comes back," because it removes one of the conditions of possibility from the historical trajectory. The book calls this configuration the final paradox and draws argumentative consequences, fully developing what can only be hinted at within a single column piece.
From that paradox descends the central concept of the book, Ghost GDP. It describes the temporal disconnection between the productivity gains that artificial intelligence brings to corporate balance sheets and the social costs that ensue. Profits appear within quarters, in the form of higher margins, compressed operating costs, and job reductions instead of layoffs. Social costs follow a completely different dynamic, consisting of intermediate roles becoming empty for years, entry-level pathways shortening, professional transitions becoming longer and more costly, entire cohorts of workers exiting the market without adequate reskilling counterparts. For a period that could last a long time, GDP and balance sheets tell a positive story, while beneath the surface, a social debt forms that isn’t explicitly accounted for. When this latency runs out, the bill arrives all at once, and at that point, the transition policies that should have been built earlier are absent.
The numbers are already speaking now. The International Monetary Fund estimates that in advanced economies, 60% of jobs are exposed to the impact of AI. The World Economic Forum, in its Future of Jobs Report 2025, predicts that by 2030, 92 million jobs will be eliminated worldwide, while 170 million new ones will be created. The net arithmetic, taken as it is, suggests that everything is fine, but those eliminated jobs and newly created jobs are not the same thing; they require different skills, pay different wages, and exist in different places. The Global AI Jobs Barometer 2025 by PwC, which analyzes nearly one billion job postings across six continents, calculated that between 2018 and 2024, productivity growth in sectors most exposed to AI almost quadrupled, from 7% to 27%. In less exposed sectors, productivity remained largely stagnant. This is precisely the asymmetry that Ghost GDP describes.
On the Italian side, the data tells a slightly different, yet equally serious, story. According to ISTAT, the adoption of AI in companies with at least ten employees has doubled in just one year, from 8.2% in 2024 to 16.4% in 2025. This means that the main shockwave, in a production system dominated by small and medium enterprises, has yet to arrive for the vast majority of companies, while for the minority currently experiencing it, the adoption curve is steep.
Bank of Italy reports that 17.4% of companies adopting AI foresee a reduction in staff, compared to 2% anticipating an increase. The remaining 70% claim that nothing will change. I leave it to the reader's discretion to decide how much of that 70% is realism, how much is inertia, and how much is a form of polite dismissal.
The second part of the book is called The Meaning and does a job that the column cannot do by nature of the format. It shifts the analysis from the economic to the existential plane and investigates what happens when work, which has been the main provider of identity for adult Westerners for a couple of centuries, stops performing that function. Within this section, you find cognitive offloading, or the externalization of thought to machines, which seems convenient until you realize it is atrophying the very cognitive skills on which your market value rested. There’s the category of "new useless", people cut off from the qualified job market at an age where serious re-skilling is materially difficult, and where the difference between the theory of lifelong learning and concrete experiences becomes visibly apparent. There’s also, particularly, an attempt to map how to construct meaning in places that employment no longer supplies. The background readings include Weber, Sennett, Gorz, Frankl, Turkle, authors who provide depth and authority within a book.
The third part of the book is called The Answer and attempts to provide guidance on three levels: individual, organizational, institutional. A chapter is dedicated to Law 132/2025, the first comprehensive Italian regulatory framework on artificial intelligence, which came into effect on October 10, 2025, transposing the European AI Act. The book analyzes in detail the articles that directly pertain to the relationship between AI and work. I acknowledge the law's real merits, from transparency to the centrality of the individual to the obligation to inform workers about AI systems used at their workplaces. However, I also note what I believe is a structural limit. The law focuses on how AI is used within companies and does not address the macroeconomic consequences of widespread adoption, nor does it foresee rapid intervention tools for professional re-skilling at an adequate scale.
One thing you won’t find in the book is a final chapter of reassuring solutions. I’m not convinced I have any, and it didn’t seem honest to pretend otherwise. What you will find, instead, is the voice of someone who is navigating this transition while telling it.
It’s an experience that concerns many people of my generation, and many are hesitant to share it aloud, for understandable reasons related to job markets, careers, and the networks of relationships involved.
Telling this professional moment while experiencing it is an uncomfortable exercise, and I’ve done it because expressing doubts and uncertainties is, in my opinion, a cathartic, liberating moment that helps in making even difficult decisions which can turn out positive in the long run.
This somewhat anomalous position, of an insider in the tech sector, a senior in transition, a concerned Italian citizen reflecting on his country, is the perspective that I felt was missing in our domestic debate, which is usually polarized between academics explaining from the lectern and startup founders selling solutions.
IA Upgrade has also been a space for exchange during these months. The questions I have received from readers, comments on articles, and testimonies from those sharing their experiences have fed the book's themes more than I initially suspected. Several chapters were born from conversations that began here and progressed thanks to the concrete stories of people doing very different jobs who ask the same questions.
I owe much to you, readers, and thanking you, I dedicate this book to you.
Note: Humans for Now -- Why This Time Work Isn’t Coming Back is available on Amazon in paperback, hardcover, and Kindle (ISBN 9798251736472). Materials, notes, excerpts, and updates at ballerano.com/umani-per-ora.