# Cognitive Atrophy **Domain:** Cognition / Artificial Intelligence / Human Development **Doc Type:** Concept Node **Maturity:** Developing ## Definition **Cognitive atrophy** is the loss or weakening of intellectual capacities through persistent nonuse, especially when remembering, reasoning, interpreting, planning, or evaluating are repeatedly delegated to external systems. The term names a functional risk rather than a claim that every use of artificial intelligence diminishes cognition. ## Mechanism Tools can remove friction while either preserving or replacing learning. An AI that explains a subject, generates counterexamples, tests a model, or helps a person practice can function as a [[wiki/Cognitive Prosthesis|cognitive prosthesis]]. An AI that supplies decisions and conclusions without requiring understanding can reduce the occasions on which the user constructs mental models, retrieves knowledge, tolerates uncertainty, and detects error. ## Protective Practice [[wiki/Cognitive Agency|Cognitive agency]] supplies the countermeasure: use AI to deepen understanding, preserve the ability to teach and apply the subject, inspect evidence, and keep consequential judgment contestable. The relevant distinction is not human-only versus machine-assisted thought. It is augmentation that leaves the user more capable versus outsourcing that leaves the user increasingly unable to act without the system. ## Relationships [[wiki/Cognitive Agency|Cognitive Agency]] · [[wiki/Cognitive Prosthesis|Cognitive Prosthesis]] · [[wiki/Human Agency|Human Agency]] · [[wiki/Cognitive Sovereignty|Cognitive Sovereignty]] · [[wiki/AI-Mediated Communication|AI-Mediated Communication]] ## Simple Reminders, Quotations, and Thoughts > “If we really believe in a humanist AI, then that means that an AI should always be trying to make humans smarter and more capable, not use humans to make itself smarter.” > **— Mustafa Suleyman**, *Humanist AI Code of Conduct discussion, September 2026* [[reminders/Cognitive Agency/Humanist AI Should Make Humans Smarter, Not Use Them by Mustafa Suleyman|Humanist AI Should Make Humans Smarter, Not Use Them by Mustafa Suleyman]] > I can feel the pull of cognitive outsourcing in my own life, so I have imposed a rule: I never use AI merely to obtain an answer; I use AI to understand a topic. If I understand the subject well enough to teach and deploy it, then AI can accelerate me without replacing my thinking. > **— Adapted from Tom Bilyeu**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Cognitive Agency/Use AI to Understand Rather Than Replace Your Thinking by Tom Bilyeu|Use AI to Understand Rather Than Replace Your Thinking by Tom Bilyeu]] ## Sources / Provenance - [[research/AI Is Now So Good They Will Pay You to Quit|AI Is Now So Good They Will Pay You to Quit]]. - [[articles/Epistemic Invariance in Conversational AI|Epistemic Invariance in Conversational AI]].