All rise for JudgeGPT

Bridget McCormack, former Michigan Supreme Court Chief Justice, now leads the American Arbitration Association (AAA), which has developed an AI Arbitrator using OpenAI models to help settle document-based disputes affordably and quickly. This system only handles cases relying solely on documents, and crucially, maintains a human in the loop at every stage, including the final award issuance. While generative AI has faced scrutiny in the courtroom due to issues like hallucination and bias, proponents like McCormack see massive potential for increasing access to justice for small and medium businesses that cannot afford traditional legal help.

Courts are already using AI for administrative tasks, and some judges, like Kevin Newsom, are exploring its use for interpreting the ordinary meaning of words in legal texts, though academics warn that LLMs are not reliable lookup engines for language and can import biases or foreign legal concepts. Concerns persist regarding AI's tendency to hallucinate facts, which poses the greatest risk to under-resourced litigants, even in specialized legal tools like LexisNexis and Westlaw that use retrieval-augmented generation (RAG).

The AAA's AI Arbitrator aims to resolve construction document disputes in 30-45 days, a significant reduction from the 60-75 days typical for human arbitrators, projecting at least a 35% cost saving. The process involves the AI summarizing submissions, organizing claims, and drafting an award, which a human arbitrator then validates, edits, and signs off on. McCormack believes this efficiency frees up human decision-makers for complex cases, while also addressing procedural justice by making parties feel heard, a factor that enhances trust in the legal process, even if they ultimately prefer human judges.

(Source：The Verge)