Abstract
This article recommends the development and deployment of an Artificial Intelligence (AI) large language model (LLM) tool by and for the federal judiciary. LLMs are revolutionizing legal practice, and judges should accordingly be familiar and in regular practice with LLM technology. LLMs can greatly assist the courts by increasing operational efficiency and public access to justice. Particularly, LLMs can improve efficiency in legal research, legal writing, and administration. LLMs present limitations or potential problems in the form of prompt dependency and harmful bias, limitations of data in their training sets, privacy and confidentiality, and the production of false or fabricated information, otherwise known as “hallucinations.” These issues, with the exception of hallucinations, may be ameliorated through the use of an internal, closed-system LLM in the federal judiciary. Negative impacts resulting from hallucinations may be limited by high user familiarity and skill with the use of LLM technology.
Recommended Citation
Nick McKinley,
Art. III-GPT,
34
Cath. U. J. L. & Tech
1
(2025).
Available at:
https://scholarship.law.edu/jlt/vol34/iss1/3
