Predicting Automation of Professional Jobs in Healthcare

Abstract

Highly-skilled professional jobs have been considered somewhat resistant to automation due to their reliance on judgement and creativity. Still, recent technological advancements such as artificial intelligence are threatening to disrupt even the jobs of professionals. This is particularly relevant in healthcare which accounts for one quarter of all professional jobs in the U.S. We test a model for predicting job automation based on concepts from recent research literature and extensive U.S. job data. We demonstrate that low automation of professional jobs can be attributed to creative skill requirements and interpersonal skill requirements. When we repeat the analysis with just the healthcare jobs we find that professional training seems to relate to lower amounts of job automation independent from creative and interpersonal skill requirements. Healthcare professions seem resistant to automation beyond what a factor model would explain. We provide theories for the unusually low automation of the jobs of healthcare professionals.