Andrej Karpathy published a structured analysis of AI's impact on labor, visualized at karpathy.ai/jobs, estimating 150 million jobs face meaningful automation risk. Financial analysts sit near the top of the endangered list. Hands-on physical roles, the kind that require spatial reasoning and fine motor work in unpredictable environments, remain the hardest for current AI systems to displace.
The episode is worth reading in full not for the list itself but for the reasoning behind it: why cognitive, sedentary, rule-bound work is more vulnerable than it looks, and how economic indicators are already showing early signals of displacement. The hosts Josh Kale and Ejaaz pull from Karpathy's framework to map which skill categories are structural liabilities versus structural advantages going forward.
The five surviving job categories hinge on physical presence, human trust, and contextual unpredictability. The conversation ends on the practical question: how workers and organizations should actually use AI now, before the displacement accelerates. Karpathy's visualizer tool is the concrete starting point for anyone who wants to stress-test their own role.
[WATCH ON YOUTUBE →]