TeamCAMS open source platform simulates human AI teamwork scenarios
TeamCAMS: An Open-Source Research Platform for Studying Human Behaviour in Human-AI Teams
Human-Computer Interaction
Summary
Managing teamwork between people and AI can be complicated. The authors created TeamCAMS, a computer simulation where researchers test how humans and AI interact while working on multiple tasks, like controlling a system. Past versions helped study effects like stress and automation reliability on performance. The newest version adds ways to explore teamwork issues like conflicts and leadership automation. TeamCAMS is freely available to encourage more research and improvements.
What this means in practice
- •For human factors engineers: Simulate realistic human-AI teamwork tasks to evaluate team performance and conflict under various automation conditions.
- •For team dynamics researchers: Study effects of distributed teamwork and leadership automation using a controllable multi-task simulation environment.
Authors
Amos Brocco, Alain Chavaillaz, Andreas Sonderegger, Juergen Sauer
Abstract
In this article, we present TeamCAMS (Cabin Air Management System), a collaborative work environment for simulating human-AI (artificial intelligence) interaction for scientific research. The article outlines how several psychological theories guided the development of this multiple-task simulation. Modelling a process control environment, previous versions of TeamCAMS have already been used in empirical studies to address a wide range of research questions (e.g., comparing different forms of automation, evaluating impact of automation reliability, effects of stress on multiple-task performance). Outlining the technical possibilities offered by TeamCAMS, the article points out how its latest version offers researchers the possibility of addressing a set of new research questions including problems associated with teamwork (e.g., within-team conflict, distributed teamwork) and human-AI interaction. Finally, we will outline how this simulation environment can be enhanced further still to address research questions in new fields (e.g., automation of leadership). To promote transparency, reproducibility, and further development, TeamCAMS is made available to the research community under an open-source license.