Teaching Justice: Reimagining STEM Education for Ethical AI Systems

Why do we keep building AI systems that harm people? Despite a decade of evidence from scholars such as Safiya Noble, Cathy O’Neil, and Ruha Benjamin documenting algorithmic bias in facial recognition, predictive policing, and automated hiring, the systems keep getting built, and the harms keep happening. This article argues that the failure is not downstream in the code but upstream in the classroom. STEM curricula still train engineers, computer scientists, and designers to ask “Does it work?” before “Should it exist?”, treating ethics as an elective, a single lecture, or a checkbox on the way to deployment. Drawing on critical pedagogy, computational rhetoric, and design justice scholarship, the paper proposes four pedagogical principles for reimagining technical education: privileging question-posing over solutionism; cultivating rhetorical awareness of AI systems as communicative actors; centering participatory and community-engaged methods; and teaching genealogical analysis of how particular technical solutions come to seem inevitable. These principles translate into concrete classroom practices, from reframed assignments to community partnerships. This article closes by arguing that pedagogical reform must be accompanied by institutional change, since practitioners cannot make ethical choices in environments that punish them for doing so.