Researchers at TU Berlin make trust dynamics in human-AI teams measurable. The results show when people outsource cognitive processes and when they retain control
Trust in artificial intelligence develops in fractions of a second – and determines whether people rely on a system or mistrustfully check its results. Researchers at Technische Universität Berlin have now shown for the first time that this dynamic can be measured directly in the brain: with the help of EEG signals. In the DFG-funded research project “Neuronal Correlates of Trust in Human-AI Interaction”, Prof. Dr. Eva Wiese and Dr. Tobias Feldmann-Wüstefeld from the Department of Cognitive Psychology and Cognitive Ergonomics are investigating how trust in AI systems is related to so-called cognitive offloading. This refers to the question of whether and when people outsource mental functions such as attention or memory to a technical system. The special feature: Instead of recording trust primarily via questionnaires or indirect behavioral data, the team relies on neuronal signals that can be measured objectively and with high temporal resolution.
