Not a day goes by without us being bombarded with clickbait news stories about Artificial Intelligence; at the same time, it is astounding how superficial the (more and more heated) debate about the topic has become. This is true for both AI’s frightened detractors and its enthusiastic supporters, who often rave about the beneficial effects of the technologies, described as the fruit of a benevolent capitalism that is the creator of a way of life that is going to make work obsolete.
What needs to be urgently emphasized is that work – its quantity and quality, its relationship with innovation and knowledge, its nexus with macroeconomic and microeconomic policies – is in danger of disappearing off the radar once again, after decades of theoretical obscurity and political invisibility brought about by neoliberalism (which came with fragmentation, precariousness, non-traditional forms). This would have terrible consequences, particularly for young people and women.
To keep work firmly in the spotlight, two things are necessary.
The first is to have a serious analytical grounding for the issue of Artificial Intelligence (the technology behind ChatGPT is only one of its forms, based on large language models that perform better than machine learning and deep learning in terms of processing and prediction capabilities – a technology which managed to reach 100 million users in a very short time, something that didn’t happen with previous innovations).
The second is to pose the fundamental question of whether innovation directed towards Artificial Intelligence is the best use that can be made of it, and, more generally, whether innovation, instead of being left up to market forces (which, moreover, were late to arrive in many past innovation cycles, coming well after the impulse given by a public actor), could be directed in a “higher” direction: for example, towards more noble purposes than eliminating jobs, such as job creation and the satisfaction of unmet social needs. This was, after all, the great issue that preoccupied Keynes the most: “The enormous anomaly of unemployment in a world full of wants.”
It’s not just a matter of controlling technologies after the fact, to mitigate and avoid their distortions, dangers, manipulative uses, disruptive failures and violations of privacy (although these in themselves are huge tasks). It’s a matter of imagining and devising ex ante new technologies and innovation cycles, alternative to those dominated by the big corporations, as Anthony Atkinson suggested (and on which we will begin to work in a June 14 seminar sponsored by La Rivista della Politiche Sociali).
There is no such thing as innovation that is “automatically good” – not even digitization. The rhetoric that describes a phenomenon as happening naturally and being outside the realm of human agency is often used to make the case for its neutrality. But we cannot fail to see that it’s possible for innovation to be driven with explicit and determined intentionality by the public authority and social actors – as was the case with the challenge put forward by DARPA (a U.S. government agency) in 2004, which offered a million-dollar prize for a driverless car. Google’s Self-Driving Car project was a direct result of the innovation spurred by the challenge. And if such intentional “directedness” was possible for self-driving cars, why shouldn’t it be possible to spur the creation of other, more socially useful innovations, geared towards satisfying great unmet needs, starting with job creation, without dismissing this as obsolete in the coming “jobless society”?
This is the key point: even if it were true – and it’s not – that we are evolving toward a “society without work,” our responsibility is to devise, invent, imagine an alternative development model structured on “full and good employment” with which to nurture the flourishing of territories, the environment, cities, education, health, social and cultural goods, children and young people.
The challenge is to conceive of innovation and new technologies not as an unintentional, inscrutable, naturalistically-determined process, but as one that can be intentionally and strategically articulated and shaped into alternative pathways that take advantage of the great “crossroads” that we come across.
Daron Acemoglous has written about the extraordinary “crossroads” that is now before us with the evolution of Artificial Intelligence; on the one hand, he argues that it’s an erroneous presumption that the direction already taken by its advance – towards eliminating jobs and with uses geared towards facial recognition, language processing, devising algorithms to replace human cognition, rather than meeting social needs such as education, upbringing, care – is the only possible one. On the other hand, he elaborates mathematical and econometric formulations to show the plausibility and feasibility of options for an alternative “directedness.”