Megan Stevenson on Algorithmic Risk Assessment
In this episode, Megan T. Stevenson, Assistant Professor of Law at George Mason University Antonin Scalia Law School, discusses her article "Algorithmic Risk Assessment in the Hands of Humans," which she co-authored with Jennifer L. Doleac. Stevenson begins by explaining how and why courts traditionally sentenced criminal defendants, focusing on the goal of incapacitation, which requires an assessment of the risk of recidivism. She observes that algorithmic risk assessment promises to make incapacitation more efficient, but notes many potential concern. She describes the empirical study of Virginia's program she conducted with Doleac, and observes that judges seem to depart from the algorithm's recommendations, especially for young defendants. She also observes that judges unexpectedly imposed lighter sentences on low-risk sex offenders. Stevenson is on Twitter at @MeganTStevenson.