Uncovering adoption personas for generative AI in higher education: a clustering-based segmentation approach

The growing integration of Generative Artificial Intelligence (GenAI) in higher education has prompted critical questions about how diverse users perceive, interact with and adopt these technologies. This study investigates adoption personas for AI-powered chatbots by applying a person-centered approach to uncover latent user segments, moving beyond traditional variable-centered models. Using a validated dataset comprising 192 observations from multiple disciplines, the research leverages unsupervised machine learning techniques, specifically hierarchical clustering followed by k-means, to identify meaningful user subgroups based on perceptual, experiential, contextual, and demographic indicators. Data preparation involved computing composite scores for each validated construct, followed by standardization to ensure comparability across variables. Cluster validity was assessed using internal validation indices (Silhouette Score, Calinski-Harabasz Index, and Davies-Bouldin Index), with a four-cluster solution emerging as optimal. The resulting personas were labeled based on attitudinal traits and further profiled using demographic and contextual characteristics such as role, tech-savviness, study field, gender, age, and education level. The identified clusters, Cautious Achievers, Skeptical Utilitarians, Disengaged Doubters, and Engaged Enthusiasts, reflect diverse configurations of trust, usefulness, efficiency, ethical comfort, and satisfaction with AI chatbots. These personas highlight hidden heterogeneity in user attitudes and reveal how demographic and contextual factors shape adoption tendencies. The findings of the study contribute to a deeper understanding of user heterogeneity in AI adoption and offer actionable insights for the design, deployment and support of GenAI tools in higher education contexts.