Quick Buzz Feed
In a new survey, more than half of responding judges report using at least one AI tool in their judicial work
A Northwestern study surveyed federal judges across the U.S. to understand their use and outlook on artificial intelligence. The research team, led by Daniel Linna and V.S. Subrahmanian, conducted a stratified random sample survey of 502 bankruptcy, magistrate, district court, and court of appeals judges, receiving 112 responses between December 2 and 19, 2025. The survey included questions about general large language models like ChatGPT, Claude, Copilot, Gemini, Grok, and Perplexity, as well as specialized 'AI for Law' tools such as CoCounsel, Westlaw AI-Assisted, Protégé, Vincent AI, Harvey, and Legora. The findings revealed that over 60% of responding judges reported using at least one AI tool in their judicial work, though only 22.4% reported using these tools on a weekly or daily basis. Approximately 38% indicated they had never used any of the listed AI tools professionally. Judges showed a preference for 'AI for Law' tools over general-purpose AI platforms. The most common judicial use cases for AI tools were identified as conducting legal research (30%) and reviewing documents (15.5%). Similarly, judges reported that other individuals in their chambers predominantly used AI for legal research (39.8%) and document review (16.7%). The study also found a correlation between personal and professional AI use, with 38% of judges using AI daily or weekly outside of work, suggesting that personal familiarity often translates to professional adoption. However, a significant portion of judges (26.9% rarely, 25.9% never) reported little to no AI use outside of work. A critical finding was the lack of formal AI training for judges, with 45.5% stating that no such training was provided by court administration, and 15.7% being uncertain. Regarding official policies, about one-third of judges either permit (25.9%) or permit and encourage (7.4%) AI use in their chambers, while approximately 20% formally prohibit it, 17.6% discourage it, and 24.1% have no official policy. The researchers emphasize the need for training, best practices, and clear policies to guide the responsible implementation of AI in the courts, noting that judges' opinions on AI's potential for the judiciary were nearly evenly divided between optimism and concern.
The study was a collaborative effort, with contributions from several co-authors in addition to Daniel Linna and V.S. Subrahmanian. These included Anika Jaitley, an undergraduate student at Northwestern; Siyu Tao, a student at Northwestern Pritzker Law; and U.S. District Judge Xavier Rodriguez, who serves the Western District of Texas. Their combined expertise helped shape the comprehensive analysis of AI adoption among federal judges.