Tribune written in a team with Jérémie Giniaux-Kats, Erik Campanini, Mathias Dufour, Florian Forestier, Pauline Maury and Sylvie Teynier, members of the think tank and action lab #Leplusimportant. Algorithms, used for a long time to optimize the operation of machines, are also increasingly used to organize work and manage human resources.
Algorithmic management, in the process of disrupting work and social relations
This use of artificial intelligence (AI) at work, also known as “algorithmic management”, is disrupting work and relations between employers and workers. It is critical to take the measure and act now to frame it. AI is, for example, increasingly used in recruiting, for filtering applications and evaluating candidates during interviews, tests or automated analysis of videos. But also for the distribution of tasks, monitoring and evaluation of workers’ performance or to decide on their promotions or even their departure!
If these uses of AI at work are developing today in digital platforms, of course, but also in distribution, call centers and warehouses, they will become widespread in all sectors. According to a study by Mercer, in 2020 39% of HR departments use predictive algorithms, against 10% in 2016, i.e. a quadrupling in 4 years!
AI at work, a source of opportunities and threats for assets
Like any great technological innovation, AI at work is a source of both opportunities and threats for workers. On the positive side, AI has the potential to make workers’ activity more productive and easier, less subject to arbitrariness, more interesting. A study of Japanese workers shows that those who work with algorithms have a higher level of satisfaction. AI can help ensure that workers are treated in a way that is more personalized to their needs and aspirations.
Conversely, AI also feeds workers the risks of abusive surveillance, discrimination, loss of autonomy and loss of employability. The exceptional development of machines’ capacities for collecting and processing personal data on the work and life of employees constitutes the backdrop for the development of AI at work. Like “smart badges” that measure workers’ productivity, track their interactions and analyze their conversations, or AI tools that analyze emails to identify the most efficient workers or deviant behavior.
AI as the source of increased discrimination
AI could also be the bearer of increased discrimination, even if it is involuntary but linked to the opacity of the decision criteria of algorithms. Amazon thus gave up on a recruitment algorithm program, having found that it discriminated against female candidates for tech jobs. Algorithmic management will lead a massive number of workers to see their work influenced or even directed by machines, without understanding why or how. As the ILO points out, the perception of being an “appendix of the machine” can lead to a feeling of alienation.
Even among skilled workers and managers, the prospect of a transfer of decision-making capacity to the machine fuels fears of loss of autonomy. Finally, the fact that a worker does not master AI creates a risk of increased wage disparities, obsolescence of skills and reduced employability. This calls for specific measures from companies and public authorities to train workers to work with AI.
Promote “responsible algorithmic management”
It is therefore necessary to regain control of AI at work, to exploit its potential while guaranteeing working and employment conditions that preserve the dignity, health, equal treatment and autonomy of workers. Let us not be afraid of “technological determinism”: nothing is inexorable and history has shown how much regulation by public authorities and business practices can guide and supervise technological developments. It is also necessary to act on time and sufficiently hard. We therefore propose several guiding principles to promote “responsible algorithmic management”.
For companies, such as public employers, AI is a way to more effectively fulfill their missions. The constraints to be imposed on them must therefore be strictly necessary and proportionate to the aims pursued, such as the protection of the fundamental rights and freedoms of individuals. In the name of entrepreneurial freedom, employers have the right to ensure that the regulatory framework on AI at work respects the capacity for innovation and intellectual property and guarantees a fair competitive game and their legal security.
Conversely, the scale of the issues imposes specific responsibilities on them: informing workers about their uses of AI to organize work, in particular on the personal data collected, training them in AI practices at work, maintain their employability, make AI and its uses a topic of social dialogue and be open to external audits to verify compliance with these points and with workers’ rights. To prevent the risks posed by AI at work, we propose to assert specific rights for workers, relating to their right to dignity (right to privacy, right to information and consent, right to human interactions ), their right to equal treatment (right to non-discrimination), their right to health and their right to autonomy (right to understanding or explicability of algorithms, right to human decision – vs. decision by algorithm for what concerns them-, right to challenge, right to training in the use of AI and right to negotiate).
A text on the regulation of AI is currently under discussion within the European Union. After being at the forefront on data protection, France has the opportunity, as part of its EU presidency, to be on the promotion of responsible AI at work, reconciling the freedom of innovation and the promotion of workers’ rights.
–
