Automation will be a boon or a catastrophe depending on whom you listen to. This paper proposes an overlapping-generations model with endogenous school choice in which the quality of a country's education system determines how well skill supply can respond to increased demand from automation and subsequently whether automation will be beneficial or detrimental. In this sense, education quality in the model offers a bridge between the optimistic and pessimistic perspectives on automation. In testing the model's assumptions, the paper finds evidence that educational attainment, cognitive skills, and select noncognitive skills are associated with avoiding automation-prone occupations. Consistent with the model's predictions, census data indicate that countries have historically relied most on these types of occupations at middle-income status. The model and empirical findings suggest that it is middle-income countries that are most vulnerable to automation if their education systems are unable to affect cognitive and noncognitive skills sufficiently. As a result, automation may herald a much different growth model for developing countries: one in which developing these skills is central.