Social interaction with AI

Human-AI partnerships and human-human partnerships bear some resemblance. During interactions we form first, impressions, attribute behavior, form beliefs about the other. This has consequences on our relation in terms of trust, joint decision making and interaction patterns. Understanding Human-AI interaction thus needs methods from social psychology to understand how Human-AI relations are formed. By taking this perspective, research in human-centered XAI will bring more appropriate means of establishing partnerships with AI systems that allow designers of AI to calibrate trust, increase performance, and maintain safety when human users interact with AI.

Placebo effects in AI interaction

The placebo effect is well established in pharmacological studies. It refers to the acquisition of a real benefit from a sham treatment (e.g., a sugar pill) compared to not taking treatment. Here, the expectations of patients play a critical role in the genesis of the placebo-effect, i.e. an expectation of recovery supports recovery. Technological innovations are likewise accompanied by high expectations in users. Do placebo effects occur in studies evaluating technological innovations? This line of research aims to investigate whether the presence of a novel technical system causes higher user expectations that lead to a better evaluation of the innovative technological system.