When asked if artificial intelligence (AI) will destroy our jobs, GPT-3 remains cautious: “This question is very controversial and largely depends on the applications of artificial intelligence.” For a more detailed, and above all more interesting answer, we will rather turn to the humans who have been working on these issues for a long time, In 2013, a study published by the University of Oxford sowed a wave of panic. After analyzing 702 professions, its authors concluded that 47% of jobs in the United States would be automatable within twenty years, thanks to robots, software or artificial intelligence. Among the least risky professions were social service assistants, orthoprosthetists and stomatologists. Among the most at risk, telemarketers, watch repairers or library agents.
Almost ten years after these fatalistic and sometimes criticized predictions, economists and labor sociologists are still trying to estimate the consequences of the generalization of artificial intelligence in our lives. Some of these works break our clichés. In 2020, Michael Webb, an economist at Stanford University, surmised that unlike robots and software, which can compete with lower-skilled occupations, artificial intelligence is more of a threat to high-skilled jobs because it doesn’t just perform repetitive tasks. Older workers, who have accumulated the most experience and who adapt less quickly, would be particularly vulnerable to these upheavals. Furthermore, in 2019, the OECD gave a less dramatic estimate that economists at Oxford University: “Only 14% of existing jobs are at risk of full automation, not nearly 50% as other research suggests.” Furthermore, 32% of jobs could “deeply change”, without disappearing.
It is therefore difficult to estimate the real impact of artificial intelligence on the future job market, between the pessimists, who fear a massive explosion in unemployment, and the optimists, who believe in creative destruction (speech largely fueled by the digital industry) or the advent of a workless society. However, some specialists argue that we are on the wrong subject. “It’s not artificial intelligence that threatens employment, it’s capitalism and the frantic race for hyperprofits”, slice Antonio Casilli, professor of sociology at the Polytechnic Institute of Paris. “Investors are looking to reduce the cost of payroll through different methods. For example, by laying off en masse and then rehiring people on a freelance basis so as not to pay social security contributions. We are also witnessing the fragmentation of professions which, before, were professionalizing and linked to strong skills. Now they are separate microtasks. What we see today is less mass unemployment than a situation of generalized precariousness.
Aggravated social inequalities
In his essay Waiting for robots (Seuil, 2019)the researcher precisely tackles the “haunting prophecy” of the end of work caused by machines. He points out that artificial intelligences are devices that require a lot of human labor to operate, beyond the computer developers who create them. That’s what we call the “digital labour”. Activities (paid or unpaid) designed to enrich digital platforms, feed them with data, and therefore give them value: Deliveroo delivery man, click worker (who does small repetitive tasks such as sorting or annotating a database) or even… Internet user who has fun chatting with GPT-3, thus training the AI to improve its performance for free. It is work that is not considered work, without protection or recognition. However, this phenomenon of digital labor is accelerating, particularly with the health crisis and even more so in developing countries. Artificial intelligence is therefore already aggravating economic and social inequalities, but not necessarily as we believe. “In this debate about artificial intelligence and jobs, we have to ask ourselves this question: when we talk about automation, what are we really talking about? asks Antonio Casilli. Often, this means replacing visible people with invisible others, who are separated from the rest of the world by a screen.