Let AI Be…

For decades, the theory was that mechanization, robotization, digitalization, and so on would protect people from hard and excessively routine work. In these visions, shared by both capitalist and authoritarian societies throughout recent history, everyone stood to gain: the state, companies, and workers.

And since modernity was feverishly transforming the surrounding natural and societal environment, the story made sense and no one seemed to be in danger. Those engaged in intellectual work were expected to focus on supervising processes, checking, thinking about data, and making decisions, while those doing manual labor were expected to be upgraded into workers operating a machine—whether mechanical or digital—or supervising such industrial machinery. Because the end of work never seemed to arrive, the world hoped for work that would be easier, more creative, more dignified, more meaningful, and less dangerous.

As the specter of the Second World War recedes, the need for that kind of massive reconstruction also comes to an end. After 1970, the general optimism fades, and the dream of near-full employment proves to have been a strictly contextual reality on both sides of the Iron Curtain. The progress of automation helps and does not help at the same time. Some professions expand enormously, while others disappear. Then, gradually, computers emerge, opening the way toward what we now call artificial intelligence. Many of these developments were, of course, interestingly described in the highly fashionable science-fiction literature of the time. With the emergence of computers, the focus shifts from mechanically operating a machine to the fact that this helpful tool of our worker begins to think. At first, strictly guided; later, increasingly autonomous, offering alternatives and operating on vast quantities of data that the person behind it could not actually manage as well or within a reasonable time using only their mind, pen, and paper. Operating a computer is no longer something learned through a short vocational course, on-the-job experience, or two or three days of training at a factory. We are now talking about a profession, an intellectual occupation, years of education and experience, and practically learning a parallel reality capable of creating things from a black box equipped with keys. From there came increasingly powerful computers, laptops, the internet, smartphones, tablets, and automobiles that are more computers than cars—and the AI revolution, which is moving at a truly dizzying pace.

The problem is not merely whether we want AI or not, or whether we are capable of creating rules for integrating it into life, education, and work. The problem is whether we are capable of understanding what we use, and whether it will become so complex that understanding it will be impossible without formal education. I fear that the reasonable answer—to use it, but as a tool and nothing else; not to abuse it, and so on—is operationally nonsense.

Then these systems evolve so extensively and become so deeply embedded in everyday life—at present, their managers and users are not being held particularly accountable, because, well, we are not going to oppose progress—that they develop so quickly that whatever we regulate about them today will run up against the fact that, by the time we finish regulating them, they will no longer be what they were when regulation began.