Building capacity for AI-powered learning

A collaborative trial of conversational agents in higher education

Abstract

Generative AI-powered conversational agents represent new opportunities for personalised learning in higher education, but institutional adoption requires careful support and professional development. This paper presents a case study of a collaborative trial involving twenty-six university staff who created and tested learning agents using the Cogniti platform over six months in 2024. Participants developed twenty agents across multiple disciplines, with nine successfully deployed to teaching practice. Survey results showed exceptional satisfaction rates, with all participants rating the platform as easy to use and 95% willing to recommend it to colleagues. Qualitative analysis identified benefits including scalable personalised feedback, enhanced student engagement, and flexible learning scenarios that would otherwise require significant staff time and resources. Participants emphasised that success depends heavily on developing prompting skills and understanding AI limitations. Critical success factors included expert support from learning designers acting as 'critical friends' and opportunities for interdisciplinary learning. Results suggest that well-designed collaborative trials can overcome barriers to AI adoption in higher education by providing necessary training, support, and community while ensuring pedagogically sound implementation that maintains quality learning experiences.