Project description

Artificial agents are a topic in the digital world that can no longer be ignored. One concern of many people is that these agents no longer make comprehensible decisions. This hinders collaborative work, which is supposed to be facilitated by automation. Humans are able to know what other humans know and can adapt their actions accordingly. Whether this ability also works for artificial agents is an open question.

In this project, we will investigate which cognitive abilities allow humans to cooperate with autonomous agents. Such knowledge will help to identify how artificial agents and the collaboration with them can be better designed. To this end, an experimental paradigm is developed that allows humans and artificial agents to solve a problem together. It is observed how the human improves in anticipating the agent's actions. Also, the assumption that a better understanding of a cooperation partner leads to a better cooperation performance is tested. In cooperation with the method center in Tübingen, the goal is to model such complex cognitive processes as accurately as possible.

Cooperations

Prof. Dr. Holger Brandt, University of Tübingen

Prof. Dr. Augustin Kelava, University of Tübingen