On May 19 and May 20 a two- day conference featured highly compelling presentations on the topic of: "Algorithms, Automation and Work - How AI can improve our work and daily lives." Among the speakers was the Institutes Director, Professor Georg Borges. In the panel "Black-Box AI - How does AI become controllable?" he showed that AI can bring great benefits if applied appropriately.
Whether in candidate selection, robotic mowers or the development of self-driving cars, scientists and researchers are working to demonstrate and harness the benefits of AI in different areas of life, while also protecting society from its disadvantages or potential missteps. In search of development criteria for standards, tools, guidelines and institutions, experts from research institutes, trade unions, OECD, AI application-oriented companies and the administration discussed their ideas throughout the course of the online conference. The speakers pointed out issues, but also presented promising findings.
The main topic of the panel "Black-Box AI - How can AI be controlled?", moderated by Nikolas Becker from the Society for Computer Science, was the critical area of candidate selection in application processes. For a long time algorithms have been used in large companies to select applicants. But are those decision transparent and accountable?
In a discussion with Katharina Zweig, professor of computer science at TU Kaiserslautern, and Leonie Beining from the Foundation for New Responsibility (SNV), Professor Borges emphasized that most employers often do not fully understand what their selection system is capable of, thus making transparency a key factor that must be established beforehand. In his opinion, an independent and reasonable testing metric is an essential requirement for these selection systems. He also mentioned the research project ExamAI , which is investigating what control and test procedures for such AI systems could look like in application fields.
The goal of the participants is: We want to make a contribution to fair AI. Once you can analyze the system, it will also be possible to understand a decision, an advantage AI usage in candidate selection. After all, you can't look into the heads of human decision-makers, but you can look into machines.
The event was organized by the German Federal Ministry of Labor and Social Affairs (BMAS) and the Denkfabrik Digitale Arbeitsgesellschaft.
The programme can be found here.