Deskilling from AI use

People who lean on AI complete the task but learn or keep less of the skill, so fewer people can check AI work later.

Facts

Lane: Interfaces and society
Wave: 8 of 12 (build order, not a date)
Needs: 17 boxes all the way back through hard links
Unlocks: 0 boxes downstream through hard links
Cluster: Science and medicine
Found by: side effects lens

Evidence as of October 2026

In Anthropic's RCT of 52 developers (29 Jan 2026) AI-assisted learners scored 50% against 67% for hand-coders, and in a Lancet study (Aug 2025) endoscopists' unaided adenoma detection fell from 28% to 22% after three months with AI.

Around it

Needs · 2

Coding agents as productsstage · Built today
Clinical AI in practicecapability · Emerging

Unlocks · 0

Nothing further on the map.

Brakes

This box works against

Expertise pipelinecapability · Speculative
Code review at scalecapability · Emerging
Scalable oversightcapability · Emerging
Output verification at scalecapability · Speculative

Loops and off-ramps

Expertise pipelineharmful loop
Capex pullbackoff-ramp · cut off
Taiwan chip disruptionoff-ramp · cut off
Power shortfalloff-ramp · cut off
Capability plateauoff-ramp · cut off
Training data ruled outoff-ramp · slowed

Openings researchers' judgement, October 2026 scouting

Startup room 46/100 · rank 46 of 124 Research room 59/100 · rank 35 of 124

Crowding1.5 Lab overlap3 Small-team fit4 Buyer today2.5 Research openness4.5

Depth: light
Players: Anthropic (RCT, learning modes, Academy); OpenAI Study Mode; Scholé AI ($3M, Feb 2026); academic work (MIT Caosun and Aral; Lancet endoscopy)
Wedge: Longitudinal studies of unaided-skill retention under agentic tools, plus an AI-off competence check firms can embed in their workflows; a product once regulators ask for proof. research