Essay proposes ‘work vs. gym’ test to decide when AI assistance is appropriate

Bruce Schneier, a public policy lecturer at Harvard Kennedy School and the University of Toronto, argues in an essay that AI assistance should be evaluated through a ‘work vs. gym’ framework. In this framing, ‘work’ refers to tasks where the outcome is the sole priority and the process is irrelevant, making AI assistance appropriate. ‘Gym,’ by contrast, refers to tasks where the process itself is the point—such as learning to write or developing critical thinking—making AI assistance counterproductive.

Schneier applies this framework to education, where he states that writing assignments are ‘gym’ tasks designed to develop skills like argumentation, drafting, and revision. He warns that students who use AI to complete these assignments risk undermining their own skill development, as the discomfort of grappling with ideas in prose is an essential part of the learning process.

The essay also extends the framework to creative fields, arguing that much of what is traditionally labeled ‘art’ is actually ‘work’—routine, functional output like instruction manuals or corporate branding—where AI can reliably produce acceptable results. However, when the process of creation is itself valuable—such as writing a novel or composing poetry—AI assistance risks eroding the very skills it aims to augment.

Schneier acknowledges the practical pressures students face, including competition and institutional incentives, which push them toward AI use even when it conflicts with learning goals. He suggests that individuals and institutions must deliberately choose to preserve ‘gym’ tasks to maintain skill development in a world where AI can increasingly perform ‘work’ tasks.

The essay does not present empirical data but instead offers a conceptual model for evaluating AI use across domains. It reflects a broader concern about the unintended consequences of AI adoption in education and creative industries, where the preservation of human skill development may require deliberate resistance to automation.