Adopting GenAI in Higher Education, a Faculty Workshop Addressing Beliefs and Agency

This paper reports the results of an intensive faculty-development workshop on generative AI (GenAI), delivered to a community of university teachers with an established practice of peer observation and reflective teaching within the Mentor Project. Adopting a pre-post, single-group design (N = 26 at pre-test, N = 25 at post-test) and drawing on a four-dimension model of teacher engagement with GenAI (Perceptions, Acceptance, Knowledge, Practices), the study examines whether a short, intensive, 15-hour training can shift faculty beliefs about the pedagogical opportunities and risks of GenAI, moving teachers from an established Adopter profile toward a more advanced Collaborator practice. Using Wilcoxon signed-rank test, results show significant post-workshop increases, with large effect sizes, in beliefs related to the personalization of learning activities, student creativity, and the redefinition of the teacher's role in AI-supported environments, while none of the risk-related items showed a significant decrease: concerns about assessment integrity and process reliability remained stable, and privacy-related concerns increased. These results suggest that GenAI training is most effective when it extends pedagogical values faculty already hold, and cannot be separated from institutional decisions on data governance and academic-integrity standards. The paper discusses implications for the design of online faculty-training programs supporting sustainable GenAI adoption in higher education.