The Future of Employment: How Susceptible are Jobs to Computerisation?

Dr Carl Benedikt Frey (Oxford Martin School)

Dr Carl Benedikt FreyDr Carl Benedikt Frey
In this project, Frey and his team developed a novel approach to measure the exposure of jobs to automation. They did so against a background of the rapidly expanding scope of tasks computer technologies can perform, following recent developments in machine learning and various sub-fields of artificial intelligence. They began by asking in which tasks do human workers still hold the comparative advantage despite these advances in technology. To answer this question, they gathered a group of machine learning and mobile robotics experts at the Department of Engineering Sciences, to determine current boundary conditions (or engineering bottlenecks) to automation. In a second step, using detailed data on occupational descriptions, they employed machine learning algorithms to examine how intensive jobs are in tasks that correspond to the identified bottlenecks to automation, allowing them to assess the exposure of existing jobs to new technologies.

This approach provides policy makers with valuable information about i) the potential scope of automation in terms of the share of jobs that are at risk of disappearing; ii) the types of skills that the educational system needs to provide; iii) a framework for understanding some of the impacts of new technologies on labour markets going forward.