Government agencies adjudicate more cases than all federal courts combined. This volume causes significant delays and errors for veterans, welfare claimants, immigrants, and others. For example, at the Department of Veterans Affairs, veterans wait an average of seven years for their appeal to be resolved. In immigration court, asylum seekers face a form of “asylum roulette,” where the identity of the judge matters more than the facts of the case.
RegLab is working to develop the evidence and toolkit for how to improve mass adjudication. RegLab conducted the first rigorous evaluation of existing quality improvement initiatives at the Board of Veterans Appeals with the former Chief of Quality Review and found the program did not reduce errors despite the significant cost of the program. Our ongoing research is focused on whether state-of-the-art NLP can help mass adjudication decisions be made more fairly, accurately, and quickly. The goal is an adjudication system better able to help those it is designed to serve and a new model for technology to promote justice at scale.