Answers are grounded in this paper and cite their sources.
Higher education largely frames generative AI (GAI) as a productivity tool, overlooking a relational problem: how globally distributed student teams build shared understanding and trust. This paper reports on two conversational AI bots built on BoodleBox and designed as “teammates” rather than assistants: AINO, used in an in-person Erasmus+ Blended Intensive Programme (BIP) in Finland, and Collaboration Cosmo, used in the eight-week, fully virtual Cosmogogy Collaborative (CosCollab). Drawing on Cosmogogy’s Human-AI-World ecosystem, we argue that an AI teammate’s value lies as much in the relational infrastructure it holds (memory, mediation, inclusion) as in its outputs. Analysis of CosCollab chat logs identified the roles Cosmo assumed, while reflexive thematic analysis of BIP focus group and reflection data revealed four themes: expectation setting, bot dominance and the anthropomorphisation paradox, language equity, and AI as instrumental partner versus formative presence. We propose design refinements for AI-embedded global collaboration.
No takes yet. Share an insight, caveat, or question.
Lindsay et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: