| Creators: |
Huber, Stephan and Papenfuß, Nathalie B.E. and Weppert, Lea and Wohlfart, Valentina and Basch, Johannes M. and Happel, Oliver and Grundgeiger, Tobias |
| Abstract (ENG): |
Enhancing human teams with AI is currently aspired across application domains. However, the prospective team role of high-performing AI-agents is rarely considered. In this paper, we propose a new method complementing existing user-centered design processes to define an AI-agent personality that suits the team. Inspired by personnel selection methods, the hiring an AI workshop prompts teams from safety-critical domains to 1) remember past incidents, 2) envision the personality and competence of their desired teammate, and 3) validate how the teammate would act in described incidents. Clustering and modeling the data made it accessible to interdisciplinary teams. Incorporating the models into the ideation process and revisiting the data for a personality drill-down exceeded insights from contextual inquiries and had a measurable impact (i.e., downstream utility) on the visions. Participants’ humanizing descriptions of a knowing and competent, yet modest teammate raise challenges for technical implementation and questions on future team-collaboration in safety-critical domains. |
| Citation: |
Huber, Stephan and Papenfuß, Nathalie B.E. and Weppert, Lea and Wohlfart, Valentina and Basch, Johannes M. and Happel, Oliver and Grundgeiger, Tobias
(2024)
Hiring an AI : Incorporating Personnel Selection Methods in User-Centered Design to Design AI Agents for Safety-Critical Domains.
In: (Adjunct Proceedings of the 2024) 13th Nordic Conference on Human-Computer Interaction (NordiCHI), October 13 - 16, 2024, Uppsala, Sweden, Art. No. 4.
ISBN 9798400709654
|
