Bias in human-AI interaction

Decisions that are made by joint human-algorithm teams are become more mainstream. Yet, it is not fully clear how decision-making processes change when the decision is not made by a single individual, not by a team of humans, but instead by a human-algorithm team. In this project, we study how decision making behavior of humans changes when they are supported by algorithmic advice. This is done in multiple domains, including the medical domain (cf., radiologists that use computer-aided diagnosis) and the railway domain (cf., complex planning that is partly performed by a computer). Are humans biased against the advice that computers give them? This is the central question we try to answer, and in particular why this would be the case and what cognitive mechanisms are involved.

Involved researchers

Leendert van Maanen

Chris Janssen

Baptist Liefooghe

Dominik Bachmann

Talha Ozudogru