Bruce Schneier, a security technologist teaching at the Harvard Kennedy School and the Munk School at the University of Toronto, says his students regularly use AI to complete writing assignments. He argues this is a waste of tuition money, but acknowledges that if their future careers will include AI writing assistants, the temptation is understandable.
The Work vs Gym Distinction
Schneier draws on a concept from AI researcher Daniel Meissler: the difference between work and the gym. At work, if your job is moving heavy objects, you should use assistive technology like a forklift. But at the gym, having a robot lift weights for you defeats the purpose—the point is the act of lifting itself.
The same applies to any AI-capable task. If the task just needs to be done and no one cares how, AI assistance is fine. But if how the task is done matters as much as the output, using AI is counterproductive.
Writing Assignments as Gym Tasks
Schneier describes his writing assignments as gym tasks, not work tasks. He assigns policy memos not because the world needs more of them, but because the process of writing—thinking, outlining, drafting, editing, making and criticizing arguments—develops critical thinking skills. Without this mental exercise, those skills atrophy.
He notes he can still tell the difference between AI-written and student-written memos. AI-generated prose is often catchy, plausible, and grammatically perfect but not well-crafted or logically coherent. Students, lacking this discernment, see confident writing as evidence of quality ideas and use AI to bypass the discomfort of turning ideas into prose. Schneier stresses that struggling with expression is a normal, healthy stage of writing.
Impact on Creatives
The work vs gym framework also applies to creatives. Most writing jobs—instruction manuals, sales presentations, legal briefs—are work tasks suited to AI. Only occasionally is writing an art form like a book or poem, where process matters. Historically, all writing was done by humans, supporting many writers' salaries. Now, society can separate work writing from gym writing, reducing the need for human writers.
Similarly for visual artists: most commissioned images are work tasks, not art. The world needs fewer pure artworks than simple images.
Incentives and Choices
Despite explaining the distinction, Schneier says his students still use AI due to pressure, overwork, and fear of falling behind peers. There is also an incentive problem: the benefits of cognitive exercise—improved reasoning and writing—are subtle, unlike tangible gym results.
He advocates making deliberate choices: use stairs instead of elevators, walk instead of Uber, and wall off cognitive gym tasks from AI. For assignments, if it's a gym task, giving it to AI wastes everyone's time because no one learns or gets stronger.
Looking Ahead
Schneier concludes that AI will fundamentally change work, but not as fast as companies claim. Policy analysis will involve AI from now on, and students must reimagine learning that skill. The line between work and gym will shift as humans adapt, but for now the distinction is clear.
