The paper, posted on arXiv, starts from a simple observation: AI systems can now produce the essays, code, reports, summaries, plans, and decisions that institutions have long used to recognize competence. That makes the question "what is learning for?" harder to answer than it used to be.
The authors introduce "post-instrumental learning" as a way to think about learning when its traditional instrumental purpose—producing work that demonstrates ability—is no longer unique to humans. They also warn of "capacity dissolution," the risk that human capabilities erode when AI does the work instead of the learner.
The abstract does not lay out a full remedy, but it frames the problem as urgent and conceptual: if learning is no longer needed for the outputs that certify competence, its value must be redefined. Based on the abstract, the paper is a position piece rather than an empirical study, and it leaves open what a post-instrumental curriculum would actually look like.