What AI Pedagogy Misses: Ubuntu and the Ethics of 'We'

When universities in the United States and Europe design AI curricula, they tend to assume their frameworks travel well—that the ethical questions raised around large language models, algorithmic bias, and data privacy are universal concerns. But walk into a classroom at Mandela University in South Africa, or consider how Ghanaian educators approach AI integration, and you'll encounter a fundamentally different set of questions. The Ubuntu philosophy—'I am because we are'—doesn't just add diversity to the conversation; it reveals blind spots in how Western institutions conceptualize intelligence itself. The Mandela University AI adoption framework, developed within South Africa's historically complicated relationship with technology (one shaped by both colonial extraction and post-apartheid reconstruction), asks a question most U.S. syllabi never raise: whose intelligence is being optimized, and for whose benefit? Ubuntu ethics locates personhood in relationship—my intelligence is inseparable from the community's knowledge-keeping. This contrasts sharply with the Western liberal individual model embedded in most AI tools, where 'personalization' means optimizing a single user's productivity. When we automate assessment or deploy AI tutors without interrogating this assumption, we risk automating a philosophy that erodes communal learning traditions still vital in many African educational contexts. For faculty designing AI pedagogy, this means pushing beyond the standard 'responsible AI' checklist. Consider: does your curriculum acknowledge that some students come from epistemologies where knowledge is collectively held rather than individually acquired? Are your AI tools evaluating students against a conception of 'smart' that privileges Western academic norms? The practical move here is to build curriculum that names these tensions explicitly—inviting students to interrogate what gets lost when AI systems trained on Western data sets meet classrooms that have always valued different ways of knowing. That's not just ethical window-dressing; it's rigorous teaching. The practical implications are concrete. Faculty can redesign assignments to honor relational knowledge-keeping—group-based AI projects that center collective output over individual performance. Assessment rubrics can explicitly value contributions that center community impact. Most importantly, we can stop treating AI ethics as a universal framework and start treating it as an ongoing negotiation between different philosophical traditions. Our students—regardless of background—will be better served by educators who acknowledge that the AI future isn't being built by machines alone, but by the assumptions we bring to the classroom.