The more AI helps us learn, the more we lose the actual skills we need to build AI in the first place.
This growth model identifies a 'skill paradox': as AI automates the tasks used in learning-by-doing, it destroys the expertise required to supervise the AI. Eventually, the economy may collapse because there are no longer enough high-ability humans who understand the systems they are supposed to be managing.
The Ironies of Automation: A Growth Theory Perspective
SSRN · 6307943
How does the interaction between AI and human capital accumulation affect economic growth? I study this question in an endogenous growth model where human capital accumulates through learning-by-doing. AI augments human learning but also automates the tasks through which learning occurs. Since AI itself needs human capital supervision, the ironies appear: automation erodes the very skills that AI development demands. I show that if AI augments learning only modestly, the economy converges to a c