Abstract: This paper studies the effects of implementing a recommender system in the context of the Chilean School Choice System. We develop an artificial intelligence-based algorithm for suggesting schools to students. Using these suggestions as input, and conjecturing different levels of acceptance rates by the population, we evaluate the general equilibrium effects of this policy. If, on average, students accepted one suggestion each, this technology could decrease the percentage of non-assigned students by 1.5pp. However, since good schools are a scarce resource, not everyone benefits from this policy. We find minor effects on commuting distances and assigned schools’
SIMCE scores. Also, we show that this technology has small but negative impacts on social welfare from a utilitarian perspective. Our results offer powerful insight for public policy and suggest that the impacts of a recommender system in a context of rival goods may be counterintuitive.