Leggi lo stesso articolo in italiano
Perhaps due to some recent readings, or perhaps simply because my mind, over the years, has settled on the edge of the tangent, when I think about today’s reality of AI—made up of generative tools that span the entirety of human knowledge—I am reminded of the epic story of the first industrial revolution.
It was a period of profound economic and technological transformation that began in England at the end of the 18th century and later spread across Europe and the world. For the purposes of these reflections, we can leave aside the social aspects—such as the mass migration from rural areas to cities and the early stirrings of the so-called industrial system (factories, shifts). The most striking feature of this period was the introduction of machines into factories, particularly machines that were, to a certain extent, autonomous and constant, no longer dependent on variable factors such as animal power or water pressure.
Rightly or wrongly, I see a clear parallel with today’s generative AI.
This leads me to wonder whether the near future of today’s AI might already be written in the history of the steam engine. The purpose of these notes is precisely to trace the historical aspects of the steam engine and how it forcefully entered both work and society, changing history.
The Steam Engine
The key to the first industrial revolution was undoubtedly the steam engine, in the version developed by James Watt in the second half of the 18th century. Using steam to generate mechanical energy capable of moving weights and quantities had long been known. The first functioning steam engine was built by Thomas Newcomen at the beginning of the 18th century. However, it consumed a lot of coal, was slow, and had limited power. Watt’s version not only became less greedy and more powerful but also capable of producing energy quickly and, above all, in a stable and constant manner—exactly what was needed!
In itself, the steam engine was just a bulky assembly of a boiler, cylinder, piston, and connecting rod, with a fifth element—the condenser—acting as the keystone. The condenser was the crucial aspect of Watt’s innovation. It was a cold container separate from the cylinder, where the used steam could be cooled and turned back into water without cooling the cylinder, thereby saving the time and fuel that would have been needed to reheat it. However, as efficient as it was, the steam engine alone was still just a tangle of components.
The Evolution of the Textile Industry
Before the industrial revolution, textile production was family-based. Yarn was spun and fabric woven at home on handlooms and spinning wheels. The delivery of raw materials (wool and cotton) and the collection of finished fabrics was handled by what we would today call couriers, serving merchants. This system had worked for centuries but suddenly became far too slow to meet the growing demand for increasingly sophisticated fabrics and clothing.
In the 18th century, new machines appeared, such as Cartwright’s mechanical loom. These were indeed machines, but still powered by human labor or, at most, by river water and water wheels. Work slowed during droughts, and factories had to be located near powerful water sources. Watt’s engine changed everything: factories could now be built anywhere, were more powerful, never stopped, and—most importantly—enabled what we would now call scalability. That is, a single sufficiently powerful engine could drive multiple textile machines simultaneously.
Impact on Employment
Watt’s invention triggered a radical transformation of labor within a few years: shift work, fixed wages (no more piecework), and continuous production. In a sense, the typical worker moved from being autonomous to being an employee, with fixed hours, a salary, and a boss to answer to.
This was not a matter of choice but an inevitable systemic evolution.
Working for a company rather than oneself led to the disappearance of many traditional trades, including spinners and weavers. Moreover, not all those who had made a living from spinning at home could be absorbed by the new factories—either because the demand for labor was quantitatively insufficient or because some refused to “make peace with the demon of the machines.”
In this context, the Luddite movement was born, lived intensely, and then disappeared—a violent protest against machines, exploitation, and the loss of dignified work. It lasted about five years, with violent agitations and destruction of factory machines, before being harshly suppressed by the British government with martial law, executions, and deportations to Australia. Ultimately, Luddism was brief but strikingly symbolic: a few years of protests around 1815 that both embodied the legendary clash between man and machine and ignited the rise of working-class consciousness, paving the way for unions and labor movements to come.
Another major consequence of the industrial revolution was the opening of the workforce to women and children, as physical strength became a far less critical factor.
The initial effect of the machine revolution was a stagnation—or even a decrease—in employment. While new factory jobs were created, many traditional workers lost their jobs—the very spark that ignited Luddism. Moreover, the new jobs were mostly low-cost labor and often involved harsh exploitation.
Over the long term, however, the picture improved. Many new professions emerged (mechanics, technicians, engine operators, supervisors, clerks), production increased, and transport and commerce grew alongside it. Ultimately, employment even rose, though in entirely different forms and after a turbulent phase that claimed its share of victims. Worker protections took longer to establish. From the steam engine to child labor laws, regulated working hours, and safety measures, several decades passed.
Beyond the Steam Engine
So far, this is history. Looking at AI through this lens, it is relatively easy to anticipate some employment turbulence but equally clear that a major transformation of work and the creation of new professions will occur. The point I want to make, however, is not about employment impact (even in software development, where “vibe coding” might be likened to the steam engine), but about the dynamic by which a mere technological innovation unleashed the industrial revolution.
Here the parallel is between the steam engine and generative AI.
In the first industrial revolution, James Watt provided only the technical innovation. Without the commercial and organizational work of Matthew Boulton (Watt’s partner), the economic system that allowed the steam engine to spread across England and then Europe would never have emerged.
The steam engine was a concrete, standardizable technology, and the same solution could be replicated across many factories with immediate and measurable productivity gains. Can the same be said today for AI? My answer: not entirely, or at least not yet.
Today, AI follows a different path. Companies like OpenAI and ElevenLabs provide models and platforms, but for a business facing the market daily, it is just a raw engine, not a ready-made solution. A raw engine may help workers quickly—and seemingly effectively—complete daily tasks, but it falls short of a company’s broader needs. Even major solution providers, like SAP or even Deloitte, cover only intermediate levels, integrating AI into management systems in a general, though not superficial, way—but never deeply contextualized for each organization.
Unlike 19th-century textiles, today there are no business models based on AI that are replicable on a large scale. There is no single entity—like Boulton & Watt—that produces and sells the same machine everywhere. Every company has its own processes, data, and culture. Today, there is a missing layer of the “last mile” between technology and organization—the practical translation of AI into everyday work. Or, to put it more dramatically, what gives the perception of actual intelligence in AI-based solutions.
The adoption of AI in companies is decentralized, molecular, consisting of micro-local adaptations—projects of tens or perhaps hundreds of thousands of euros implemented with agility and intellectual honesty. The real revolution of AI lies not in the technology itself but in organization and vision: what can we do today with these tools to work better and achieve more? AI is only the engine; each company is responsible for building its own vehicle.
Here, in the concept of molecular diffusion, lies space for small and agile companies to emerge, grow, and replicate. Here, too, large companies and organizations must find aggregates that extend or replace existing ones, leading to rethought—or simply different—flows. After all, every true technological revolution—and in IT, we have experienced at least two recently, Internet and mobile, not to mention cloud—carries with it a breakthrough: making possible what previously simply was not.
Looking Ahead
The enforcement of the AI Act requires giving AI in business processes a structured and organic dimension. Companies are called not just to introduce AI tools but to integrate them into their information systems and define usage, control, and accountability in line with the European regulatory framework.
To achieve this, executives must understand that the AI continuum must be disaggregated—no longer seen as a magic box but broken down into a conglomerate of molecular components that together build the last mile between today’s business reality and that of the next decade.
In other words, what’s needed are smart, tailored solutions—not just one-off pieces of technology (which, ironically, are mass-produced by major players).
Pragmatic Opinions 