"Would You Want an AI Tutor?" Understanding Stakeholder Perceptions of LLM-based Systems in the Classroom

Different stakeholders have fundamentally different concerns about AI tutors: The paper demonstrates that the same LLM-based tool raises different concerns depending on whether you're a teacher, parent, student, or administrator—meaning one-size-fits-all adoption policies miss critical perspectives that should inform decision-making.

Current adoption decisions often bypass systematic stakeholder input: The research finds that schools and universities frequently deploy LLM tools without systematically engaging the full range of people affected by them, suggesting a gap between how these decisions are made and how they should be made for responsible implementation.

Understanding perceptions requires context, not just measuring approval: Rather than simply asking "do you accept this technology?", the Co-PALE framework shows that responsible AI governance requires understanding whose concerns surface, in which educational contexts, and what those concerns mean for how systems should be designed and governed.

Tensions and uncertainties exist even among willing stakeholders: Focus groups with university faculty and K-12 parents revealed that even people open to AI tutors have unresolved questions and competing priorities, indicating that stakeholder engagement should surface these tensions rather than resolve them prematurely.

Responsible AI adoption requires deliberate, context-specific decision-making: The paper argues for moving away from blanket bans or approvals toward more thoughtful governance that considers educational context and connects stakeholder concerns directly to design and deployment choices.