How will people become experts if AI performs the apprenticeship tasks through which expertise was previously developed?
Beane's 'shadow learning' research showed automation strips juniors of deliberate practice; generative AI compresses novice-expert performance gaps at work — while its effect on the practice through which expertise forms is the open question.
View on the map → · Open in Browse →
Deming supplied both halves of the question: experimental evidence that unguarded AI use short-circuits learning, and an argument that sample-efficient social learning is the durable human advantage — implying where apprenticeship still pays.
Reviews experimental evidence — including a school study where unguarded GPT-4 lowered exam scores — on cognitive offloading masking skill.
Humans' comparative advantage is our learning efficiency.
Human sample-efficient social learning as the durable expertise advantage in high-context work.