Designing Stakeholder-Based Pedagogy for AI Ethics Education: Insights from a Multi-Institutional Case Study

As AI systems increasingly shape real-world decisions, there is growing urgency to teach students not only how to build these systems but also how to consider the social impact of their design choices. One promising approach explored in computing education is the stakeholder-based pedagogy, which, in AI courses, encourages students to examine how AI development affects different groups and to design with those needs and values in mind. In this pilot study, we conducted interviews with 14 instructors across three U.S. institutions to understand how they design and implement stakeholder-based modules in undergraduate AI ethics education. Our findings reveal two core pedagogical strategies: empathy-driven role play and structured stakeholder analysis, as well as cross-cutting challenges, including the heavy labor of curating materials, limited integration of stakeholder reasoning into the broader AI curriculum, time constraints, difficulty embedding stakeholder reasoning into technical instruction, and unclear assessment methods. We also examine how these challenges are shaped by the curricular structures and institutional contexts in which the modules were implemented. Drawing on these insights, we offer recommendations to strengthen stakeholder-based AI ethics education at both the micro level (e.g., establishing shared resource repositories and assessment frameworks across institutions) and the macro level (e.g., prioritizing stakeholder reasoning as a core AI competency and fostering channels for interdisciplinary and academia-industry dialogue). This work contributes to ongoing conversations in AI ethics education on how to better prepare students to become socially responsible AI practitioners through meaningful stakeholder analysis and engagement.