Human-in-Control - Implementing employee participation in the design, development, and deployment of artificial intelligence

Artificial intelligence (AI) is leading to profound changes in the world of work. Economically and socially, it has the potential to trigger positive developments. However, its use, especially when unregulated, also poses risks to individuals and society. While the EU has taken an initial step to protect consumers with the AI Act, comparable protections are still largely lacking for employees who are required to work with AI systems.

Even modern systems inevitably make mistakes. For this reason, humans are often deployed as a control instance (“human in the loop”). In the area of high-risk systems, this will even become mandatory in the future under the AI Act. Responsibility for potential errors then lies with employees, who often lack the organizational and technical means to influence the systems.

While trade unions and other actors have been addressing for years what constitutes good work in times of digitalization, and thus increasingly in the use of AI, there is still an almost complete lack of research on how the identified measures can be concretely implemented and technically realized. Ultimately, however, it is precisely this implementation that determines what the future of work, and in particular collaboration with AI, will look like in practice.

The aim of the project is to research and develop technical methods and organizational processes that enable employees to exert influence over the design, development, and deployment of workplace AI systems, thereby safeguarding and improving job quality.