Signal When AI Does the Work, What Is Learning For? Post-Instrumental Learning and the Risk of Capacity Dissolution

Summary

A paper addresses the question of what learning is for as AI systems become capable of producing the essays, code, reports, summaries, plans, and decisions through which institutions have traditionally recognized competence. The authors note that existing AI ethics emphasizes present failures such as bias, opacity, hallucination, labor extraction, privacy risk, and weak accountability, but argue that if the case for learning rests only on those failures, each technical improvement in AI would appear to weaken that case. Using an idealization of AI that executes specified tasks flawlessly while lacking authority over purposes, legitimacy, and responsibility, they develop the concept of 'post-instrumental learning' — learning that preserves the capacities people and institutions need when many useful outputs can be delegated to AI. They identify five such capacities: end-setting, reason-giving, contestability, refusal and revision, and participation, naming their erosion 'capacity dissolution.' The central case examined is assessment under generative AI: when a polished artifact no longer reliably evidences understanding, the authors argue institutions must assess the learner's accountable relation to AI-mediated work rather than the artifact alone. The paper concludes that AI governance should evaluate not only whether systems perform well, but whether their deployment leaves people able to understand, challenge, revise, and share responsibility for the practices those systems mediate, and has been accepted at the AIES 2026 conference in Malmö, Sweden.