Student Perceptions and Preferences of an AI Learning Companion

As artificial intelligence (AI) becomes increasingly integrated into educational settings, understanding how students perceive and prefer to interact with an AI-Buddy (learning companion) is essential for responsible design. This study investigated 168 undergraduate students’ perceptions and preferences regarding an AI-Buddy, including its pedagogical role, knowledge level, identity customization, voice, mind attribution, demographic characteristics, and comfort sharing academic struggles. Results revealed that students most frequently preferred the AI-Buddy as a tutor with an omniscient or professor-level knowledge base, a non-human and neutral interface, and a warm yet robotic voice. Preferences for mind attribution were split, with nearly half preferring low mind attribution. Female students were more likely to prefer a female AI-Buddy, and male students a male AI-Buddy; White and Black students showed same-race preferences. These findings highlight the diversity of student preferences and underscore the need for a flexible, customizable AI-Buddy that prioritizes transparency and user agency rather than one-size-fits-all designs. We discuss how our findings can be extended to future research and educational AI design.