The European Commission recently presented a proposal that sets preconditions for the use of AI in managing platform workers. The Commission believes that important decisions such as who gets an assignment or loses their account can be left to AI under certain conditions. The same goes for the monitoring of platform workers. Think of delivery routes of pizza couriers, eye or mouse movements of people working on a computer, ratings by customers and so on. All this is aimed at monitoring the performance of workers, but also to evaluate them and take automated decisions about them. You cannot see these things separately.

The fundamentals of employment relationships

Catelijne Muller is concerned that these preconditions, however well-intentioned, could steer ‘algorithmic management’ in the wrong direction. “Conditions alone are not enough to manage the unavoidable impact of algorithmic management on workers’ rights, working conditions and the relationship between employers and employees. The impact should not be underestimated. After all, its concerns the fundamentals of employment relationships: monitoring, evaluation, assessment, pay, promotion, and dismissal, to name just a few. Fundamentals that have been surrounded by laws, rules and standards for many years. We must ensure that new regulation does not in fact weaken the position of the employee and undermine existing regulation.”

Algorithmic management of platform workers

Algorithmic management involves a plethora of AI practices and tools that (help) monitor, track, assess, evaluate, recruit, hire and dismiss workers. It is important to understand that it has several ‘layers’ to it. It involves tools that track workers’ activities, both on- and offline. Think of eye-tracking, the tracking of mouse movement and clicks, room noise, typing speed, but also the tracking of physical movements, such as walking speed, driving routes, etc. And it involves tools that, based on these and other inputs, take decisions about workers. These decisions can involve rostering, or the allocation of tasks, but also promotion, demotion, and even termination.

The European Commission believes that algorithmic management is acceptable if there is clear communication about it if regular evaluations take place and if certain personal data is not taken into account. Furthermore, platform workers must be made aware that AI is involved, and unions must be informed if algorithmic management is introduced or adapted in the workplace.

At the same time, the European Commission places a lot of responsibility for the use of ‘algorithmic management’ on the platform worker, instead of on the platform. The platform worker must take the initiative to request an explanation about an automatic decision for example.

Legitimisation and Techno-determinism

“What is striking, is that EU regulatory initiative for Platform Work considers ‘algorithmic management’ as a given.” says Catelijne. “As something that we can only accept. This approach, however, is more like a band aid, and in its well-intentioned effort could have the undesired effect of legitimization, normalization and mainstreaming of quite a few work-related AI applications that are still heavily scrutinized. AI of which the impact, both on employees and on existing labor legislation, is still far from fully understood. In the long run, this could even lead to the undermining of existing labor legislation and an increase in algorithmic management if the requirements for automated decision-making become less stringent than for human decision-making.”

“We should not forget that AI is not magic. It has no common sense, no understanding of context and is prone to unfair and sometimes discriminatory outcomes. It can be quite invasive, and it often leads to generalization, losing sight of individual circumstances. AI is also sensitive to ‘bias’ and it regularly comes to unfair or incomprehensible decisions. The question is whether AI is actually ready for responsible algorithmic management.” says Catelijne. “In addition, the Commission’s proposal may have an impact on the labor market as a whole. The proposal deals with platform workers but could set a trend for the role algorithmic management can play in labor relations in general.”

“Let’s not be guided by techno-determinism. We must first and foremost evaluate the legality, moral and social acceptability and desirability of these systems and the benefits they provide and for whom, especially in the employer-employee relationship. This includes guarding against weakening the position of workers and undermining existing rules and structures. Only then will AI land in the workplace in a responsible manner.” says Catelijne.