Learning with AI-animated pedagogical agents: The role of age and gender similarity in identification, situational interest, and learning outcomes

Educational videos featuring pedagogical agents (PAs) are considered effective teaching tools. Although theories suggest that sociodemographic similarity (e.g., age and gender) between PAs and learners can enhance learning, little is known about learners' perceptions of and interactions with AI-animated PAs. Accordingly, the present study investigates how the age and gender of AI-animated PAs interact with those of learners in shaping PA identification, situational interest, and learning outcomes. Participants ( N = 218) were randomly assigned to watch one of four videos in which the PAs' age (young/old) and gender (female/male) were systematically varied. Results showed that older participants (60+) identified more strongly with the PA than younger participants (18–39). No similarity effects were found; however, mediation analyses demonstrated that identification predicted situational interest, which affected learning outcomes. These findings, relevant for the effective design of PAs, indicate age-related differences in PA perception and emphasise the roles of identification and interest in learning. • PAs' age and gender did not affect participants' identification, situational interest, or learning outcomes. • Older participants (60+) identified more with PAs than younger participants (18–39). • Identification with the PA predicted situational interest, which, in turn, predicted learning outcomes. • Identification with the PA showed a negative direct association with learning outcomes.