Cognitive debt

The long-term cost of offloading thinking to AI without practicing the underlying skill. Convenience now may erode capability over time, similar to how GPS weakened navigation skills for many drivers.

The "debt" metaphor works like financial debt: you borrow convenience today and pay back in lost capability later. If AI always writes your emails, you may gradually lose your feel for tone, structure, and audience. If AI always debugs your code, you may stop building the mental models that help you spot problems independently. This pattern is familiar from other tools (GPS did measurably weaken map-reading and spatial reasoning for many people), yet AI extends it into judgment-shaped tasks like writing, analysis, and decision-making that were previously difficult to automate.

Builder example

Builders of education, writing, and knowledge tools face a design choice at every feature: does the AI teach the skill or replace the practice that builds the skill? An AI writing tutor that shows a student how to restructure a weak paragraph builds capability. An AI that silently rewrites the paragraph for the student accumulates cognitive debt. The same tool can do either, depending on how the interaction is designed.

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

Ask for the founder's claim first, then pressure-test assumptions and counterexamples before offering any prose.

Common confusion: The early research behind this concept (including an MIT EEG study on ChatGPT-assisted writing) is preliminary and has methodological limitations. Treat cognitive debt as a useful warning framework, not as settled neuroscience.