AI Chatbot As Your Teammate
The whole is bigger than the sum of parts. In many real-world tasks, such as workplace collaborations and student course projects, not only is team effort required, but also critical to task success.
What are the characteristics of successful teams? What team characteristics can be used to predict team performance and team dynamics? Studying team characteristics is challenging, since many factors, such as individual team members’ characteristics and team personality compositions, may affect team performance and team dynamics.
To study and potentially guide teaming efforts, we have developed an AI chatbot, called INDIGO (Individual Differences for Group Optimization) on the Juji platform that can interact with a user in a one-on-one text-based chat.
INDIGO is developed to achieve three purposes. First, it replaces traditional online surveys to gather initial information from team members, such as their team preferences and expectations. Interacting with a chatbot like INDIGO is a new experience, which could help combat survey fatigue and collect higher quality information [1].
Second, INDIGO replaces a traditional personality test to automatically gauge users’ personality traits objectively. In a teaming effort, team members might provide less truthful answers in a traditional personality test to make themselves more desirable to potential teammates. To prevent faking, INDIGO can automatically infer a user’s personality traits without asking any direct, self-reported personality ratings [2]. The inferred personality traits can then be used to study team personality compositions and their effect on team performance and team dynamics.
Third, INDIGO could potentially serve as an AI teammate that can follow a team and interact with team members continuously during their team efforts. Such a teammate can help conduct longitudinal team studies, during which it will detect changing team dynamics and potentially guide team behavior based on such changes.
To test out the usefulness of INDIGO, we deployed INDIGO in an educational setting, where it was used to interact with 201 students who enrolled in a large engineering class at a university, formed 40 teams, and engaged in semester-long team projects.
The results show that students interacted with INDIGO for an extensive period of time (e.g., 60-minute chat in their interview) and offered open and honest input. INDIGO also elicited rich information from the students that allowed instructors to better understand student team preferences and perceptions. Third, specific team personality compositions based on INDIGO-inferred personality traits were able to predict team perception and team performance.
For more details on INDIGO, please check out our full paper [3]published in the 24th International Conference on Intelligent User Interfaces (IUI 19').
[1] Xiao, Z., Zhou, M. X., Liao, Q. V., Mark, G., Chi, C., Chen, W., & Yang, H. (2019). Tell Me About Yourself: Using an AI-Powered Chatbot to Conduct Conversational Surveys. arXiv preprint arXiv:1905.10700.
[2] Goldberg, L. R. (1999). A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models. Personality psychology in Europe, 7(1), 7–28.
[3] Xiao, Z., Zhou, M. X., & Fu, W. T. (2019, March). Who should be my teammates: Using a conversational agent to understand individuals and help teaming. In Proceedings of the 24th International Conference on Intelligent User Interfaces (pp. 437–447).
