Advisors, Validators, and Teachers: AI Agents for a Community of Practice on Accessibility

This study explores pedagogical responses to the fifth digital revolution, arguing that educational reasoning, not only technology, settles key ethical and epistemic questions. A design-based workshop engaged 225 Italian humanities students in creating accessible landing pages while coordinating three advisory AI agents and an external validator. This process cultivated practical judgment and care for publics who rely on accessible services. All teams achieved zero-error validation (avg. 2.7 cycles). Qualitative analysis of logs and notes revealed emergent orchestration strategies and a shift from rule-following to accountable decision-making. Framed by capability theory and an ethics of care, the findings show accessibility can bridge legal mandates and educational aims, with human-AI teaming supporting student agency.