Cognitive debt

An informal hypothesis that repeated cognitive offloading can create future costs when a person or organization still needs the displaced understanding, memory, or skill.

Cognitive debt is used to describe a mismatch between short-term output gains and later capability, comprehension, or oversight. The effect depends on the task, learner, tool design, duration, and whether the displaced skill is still required. Early AI-writing studies do not establish a universal law that assisted work weakens cognition, and automation can free attention for more valuable reasoning.

Builder example

The product question is whether users must later explain, transfer, diagnose, or perform the underlying work without the tool. A tutoring workflow may preserve retrieval and revision practice; a production workflow may automate the task and build a stronger fallback system instead. Neither interaction pattern should be imposed on every use case.

A founder asks for help with a strategy memo. The assistant writes the entire argument, recommendation through conclusion.

Decide which capability the founder must retain, and design practice or fallback tests only for that capability rather than preserving manual work by default.

Common confusion: One EEG or writing study cannot establish broad long-term cognitive decline, and lower effort during a task is not itself harm. Treat the term as a design hypothesis to test against the capability the system actually needs.