Northwestern University researchers released a first-of-its-kind random-sample survey showing that more than 60% of federal judges have used at least one AI tool in their judicial work, though daily or weekly use remains limited to just 22.4% of respondents.[1][2] The study surveyed 502 federal judges across bankruptcy, magistrate, district court, and appellate levels, receiving 112 responses between December 2-19, 2025.[2]
Who's Involved
The research was conducted by Northwestern University in collaboration with the New York City Bar Association.[3] The survey examined usage of both general-purpose AI platforms—including ChatGPT (OpenAI), Claude (Anthropic), Copilot (Microsoft), Gemini (Google), Grok (X.ai), and Perplexity—as well as legal-specific AI tools like Westlaw AI-Assisted Research (Thomson Reuters), Lexis+ AI (LexisNexis), and others.[1]
Context and Key Findings
Judges demonstrate a clear preference for legal-specific AI tools over general-purpose platforms, with Westlaw AI-Assisted Research leading at 38.4% usage and ChatGPT second at 28.6%.[3] Legal research is by far the dominant use case, with 30% of judges using AI for this purpose, followed by document review at 15.5%.[1] Notably, usage patterns vary significantly by judge type: bankruptcy judges showed the highest frequency of daily or weekly use at 32.2%, compared to just 13.9% among district court judges.[3] The correlation between personal and professional AI use suggests adoption may be influenced by judges' comfort with the technology outside the courtroom.[2]
Why It's Newsworthy
This represents the first empirical study based on a random sample of federal judges regarding AI use, providing concrete data on a critical question about how artificial intelligence is reshaping the judicial system.[2] The findings arrive as courts continue grappling with governance questions around AI reliability, bias, and appropriate applications in legal proceedings—making this baseline measurement of current adoption particularly timely for policy discussions.