“F*** You!” – An Investigation of Humanness, Frustration, and Aggression in Conversational Agent Communication
Johannes Riquel, Alfred Brendel, Fabian Hildebrandt, Maike Greve, Alan R. Dennis · Journal of the Association for Information Systems · 2021
Conversational Agents (CAs) are becoming part of our everyday lives, whether in the form of voice assistants (such as Siri or Alexa) or as chatbots (for instance, on Facebook). When looking at CAs, users frequently start to display aggressive behavior towards CAs, such as insulting it. However, why some users tend to harass CAs and how this behavior can be explained remains unclear. We conducted a two-conditions online experiment with 201 participants on the interrelation of human-like design, dissatisfaction, frustration, and aggression to address this circumstance. The results reveal that frustration drives aggression. Specifically, users with high impulsivity tend to severely insult a CA when it makes an error. To prevent such behavior, our results indicate that human-like design of CA reduces dissatisfaction, which is a driver of frustration. However, we also discovered that human-like design directly increases frustration, making it a double-edged sword that has to be further investigated.