Social Intelligence towards Human-AI Teambuilding (Student Abstract)
Keywords:Human-AI Teams, Social AI, Human-AI Interaction, Trust In AI, Human Computer Interaction
AbstractAs Artificial Intelligence (AI) continues to develop, it becomes vital to understand more of the nuances of Human-AI interactions. This study aims to uncover how developers can design AI to feel more human in a work environment where only written feedback is possible. Participants will identify a location from Google Maps. To do this successfully, participants must rely on the answers provided by their teammates, one AI and one human. The experiment will run a 2x4 de-sign where AI's responses will either be designed in a human style (high humanness) or state a one-word answer (low humanness), the latter of which is more typical in machines and AI. The reliability of the AI will either be 60% or 90%, and the human will be 30%. Participants will be given a series of questionnaires to rate their opinions of the AI and rate feelings of trust, confidence and performance throughout the study. Following this study, the aim is to identify specific design elements that allow AI to feel human and successfully appear to have social intelligence in more interactive settings.
How to Cite
Bailey, M. E., & Pollick, F. E. (2023). Social Intelligence towards Human-AI Teambuilding (Student Abstract). Proceedings of the AAAI Conference on Artificial Intelligence, 37(13), 16160-16161. https://doi.org/10.1609/aaai.v37i13.26940
AAAI Student Abstract and Poster Program