Before you ask AI for help, ask this one question

AI should handle tasks where the result matters more than the process, while people should retain tasks that build their own skills. Bruce Schneier writes for Schneier on Security that the distinction offers a straightforward way to judge when AI assistance is useful and when it may weaken human capability.

Schneier describes two categories. “Work” consists of tasks that simply need completion, such as routine documentation, technical instructions, disclosures or standard business materials. If an AI system can perform these jobs accurately, securely and with errors that humans can correct, using it can be appropriate.

“Gym” tasks are different. Their purpose is not merely to produce an output. They develop the person doing them. Schneier places student writing assignments in this category. Drafting, structuring arguments, revising and struggling to express an idea help students build critical thinking skills. Handing that process to a chatbot may produce polished text, but it removes much of the learning.

Process can be the product

The framework also applies beyond education. Many writing and design commissions serve practical needs, Schneier argues, and can increasingly be automated. Yet creative work, including fiction, poetry and other forms where human expression is central, cannot be assessed only by the finished file or image.

Schneier also warns that the test begins only after an organisation has established that an AI tool is dependable. Users need to consider accuracy, correctability and security before delegating any task.

The challenge is partly social. Students and workers may feel pressure to use AI if others do so. Schneier argues that people can nevertheless protect certain cognitive exercises, much as they might choose stairs over an elevator. His central recommendation is simple: use AI for output-driven work, but preserve tasks that strengthen judgement, reasoning and craft.