Source: The Mainichi

A feasibility project by the Supreme Court found artificial intelligence may be effective at processing large volumes of documents and summarizing arguments in civil trials. The project, carried out from January to February, involved six district court judges who tested AI on mock trial documents using services such as Gemini 3 Pro. The court identified that AI could be useful in summarizing documents and evidence and in helping judges who join trials midway catch up on earlier proceedings. The Supreme Court stated it has no intention of using AI to make legal judgments on behalf of judges. The study flagged concrete risks, including instances where AI failed to properly identify a point of contention from a newly submitted document or showed insufficient legal understanding. Participants warned that AI outputs can appear well organized on the surface, which may lead judges to become overly reliant on the technology and potentially diminish their analytical skills. Verification results concluded judges will inevitably need to verify AI-generated responses themselves and correct any errors.

The survey emphasized that relying solely on AI could lead to serious omissions because AI can overlook crucial facts and evidence. One examined case showed AI failed to identify a contentious point based on newly submitted material. The project also raised concerns about overconfidence in AI among legal professionals as a practical risk. The top court plans to further examine the feasibility of introducing AI in more practical settings, including how to handle personal information and ensure digital security. Overall, the study presented a cautious stance: AI may assist in document processing and summarization, but human verification and safeguards are necessary.