Deploying Trustworthy AI in the Courtroom: Lessons from Examining Algorithm Bias in Redistricting AI

Deploying trustworthy AI is an increasingly pressing and common concern. In a court of law, the challenges are exacerbated by the confluence of a general lack of expertise in the judiciary and the rapid speed of technological advancement. We discuss the obstacles to trustworthy AI in the courtroom through a discussion that focuses on the legal landscape surrounding electoral redistricting. We focus on two particular issues, data bias and a lack of domain knowledge, and discuss how they may lead to problematic legal decisions.

We conclude with a discussion of the separate but complementary roles of technology and human deliberation. We emphasize that political fairness is a philosophical and political concept that must be conceived of through human consensus building, a process that is distinct from algorithm development.