[Invest&Law] Court: "AI Model Under Development" Prosecutors: "Criminal Procedures to Be Digitized Starting Next Year"

2024 LegalTech AI Special Show Unveils Legal Blueprint
"AI as a Tool Supporting Judicial System, Recommending Similar Case Rulings to Courts"
New KICS Promoted by Ministry of Justice and Prosecution
"Learned from 270,000 Records Including Minutes and Judgments"

From the 26th to the 28th of last month, Judge James E. Baker of the U.S. Federal Information Protection Review Court delivered the keynote speech at the "2024 Legal Tech AI Special Exhibition (LTAS, Legal Tech & AI Show)" held at Hall A on the 1st floor of COEX, Samseong-dong, Seoul, hosted by The Law Times. Photo by The Law Times

The courts and prosecution have publicly disclosed their policy directions related to artificial intelligence (AI) for the first time. The judiciary has decided to introduce AI to improve trial procedures and is actively developing models, while the prosecution announced plans to launch a next-generation criminal justice system incorporating AI into case processing tasks. As the paradigm shifts into the ‘AI era,’ attention is focused on how far AI will influence the domestic legal sector.

At the ‘2024 Legal Tech AI Special Show (LTAS)’ held for three days from the 26th of last month at COEX in Samseong-dong, Seoul, jointly hosted by Legal Times and Messe Esang, Won Hosin, Director of Judicial Informatization at the Court Administration Office (age 52, Judicial Research and Training Institute class 28), delivered a keynote speech titled ‘Artificial Intelligence and Judicial Information Systems’ on the first day. Lee Seongbeom, Head of the Information and Communication Division at the Planning and Coordination Department of the Supreme Prosecutors’ Office (age 48, class 34), also gave a keynote speech titled ‘Utilization of Generative AI in Prosecution Case Processing Tasks.’

This was the first time that the heads responsible for informatization at the courts and prosecution publicly revealed the progress of AI utilization plans, leading to reactions at the event that it felt like seeing the first blueprint for the future of AI in the legal profession.

In his keynote speech, Director Won emphasized, “In the current situation where the judiciary needs tools to support the trial system, the introduction of AI will improve the efficiency of trial procedures and enhance accuracy in trial guidance and document preparation. It will also improve accessibility to legal services and is expected to reduce procedural and research costs.”

Director Won introduced the AI models currently being developed by the Court Administration Office for the first time while presenting the ‘Prospects for AI Introduction in the Judiciary.’ These include the ‘Similar Case Judgment Document Recommendation Model (search function development among AI utilization fields),’ an algorithm model that finds and recommends cases similar to those being handled by civil, criminal, family, administrative, and patent courts for reference by the trial courts, as well as a ‘Big Data Platform’ that integratively provides collection, storage, processing, and analysis of structured and unstructured case data for AI model development.

He also presented models planned for future development (Next Generation), such as the ‘Civil Mediation and Settlement Prediction Model,’ which analyzes complaints, answers, and preparatory documents submitted by parties to provide predictive information on the possibility of mediation or settlement before the first trial date, and the ‘Claim Document Issue Extraction Model,’ which designs an algorithm to identify issue sentences in documents and determines issue sentences based on rebuttals to facts in complaints, answers, and preparatory documents.

Director Won stressed the need to establish legal and ethical standards. He explained, “The Court Administration Office is also considering the establishment of legal and ethical standards, and the introduction of AI technology was a major agenda item at the Judicial Policy Advisory Committee launched on the 12th. Through internal and external discussions within the judiciary, we plan to establish legal and ethical standards for the introduction of AI systems and improve and develop systems accordingly.”

From the left, Won Ho-shin, Director of Judicial Informatization Office at the Court Administration, and Lee Seong-beom, Chief Prosecutor of the Information and Communication Division, Planning and Coordination Department, Supreme Prosecutors' Office. Photo by Legal News

In his keynote speech, Head Lee of the Supreme Prosecutors’ Office’s Information and Communication Division stated, “It is clear that AI is the most useful auxiliary tool for improving prosecution work efficiency. We are currently reviewing how to efficiently integrate AI into prosecution tasks and plan to promote its introduction.”

In this regard, the Ministry of Justice and the prosecution are promoting the introduction of the next-generation Criminal Justice Information System (KICS). The next-generation KICS aims for ‘complete digitization’ of all criminal procedure processes. Once ‘complete digitization’ is achieved, electronic documents, currently limited to DUI and unlicensed driving cases, will be expanded to all criminal cases. According to the Ministry of Justice’s legislative notice, the application date of the Criminal Procedure Electronic Document Act is set for June 9, 2025, and electronic documents will apply only to criminal cases initiated after this date.

AI services utilized in the next-generation KICS include similar case investigation document recommendation and voice recognition support for report preparation. The similar case investigation document recommendation service suggests investigation reports, decisions, and judgments of similar cases to prosecutors for use in determining prosecution and applicable laws. Approximately 270,000 documents of 10 types, including investigation reports, indictments, non-prosecution decisions, and judgments, were used for training. Since a big data analysis platform has been established, related documents created in new cases will be automatically learned going forward.

The voice recognition report preparation support service converts conversations between investigators and subjects into text in real time. It can be used not only for voice recognition during investigations but also for submitted video or audio files, converting the audio contained in files into text for investigative use.

Head Lee also announced the results of a commissioned study on the potential use of generative AI in prosecution work. He introduced, “We selected seven work items where AI application is possible: ▲statement summarization and analysis ▲investigation information summarization and document draft preparation ▲extraction of meaningful information from messengers and PCs ▲reconstruction of legal expression sentences ▲sentencing proposals ▲review of crime constituent elements and prosecution requirements ▲identification of missing information.”

According to the study results, AI introduction enabled rapid identification of key case contents and issues and effective information provision. However, since prosecution work uses a closed network and ‘security’ is critical, there were limitations. For example, investigation report data anonymized for security purposes lacked clear subjects, and prosecutors’ re-verification was essential for investigation document preparation, which requires specialized knowledge and experience. Regarding sentencing, since sentencing guidelines are added or changed annually, a large amount of data was needed.

Head Lee pointed out, “The introduction of generative AI has clear advantages such as work optimization, time and cost reduction, and stabilization of the judicial system. However, considering data security within the prosecution, there is also the disadvantage of high initial costs due to the need to build an internal server.”

Legal Times reporters Park Suyeon and Woo Bin