Can AI Become Part Of Judicial Reform?

This year may have already become a record-breaking one in terms of the number of statements made by officials responsible for the judicial reform regarding personnel problems in Ukrainian courts.

Here is a recent statement by the head of the High Council of Justice, Hryhoriir Usyk: “At present, the number of judges who resign far exceeds the number of candidates for judgeships that the High Qualification Commission of Judges of Ukraine is able to select through competitive procedures and the number which the High Council of Justice can recommend for appointment. Accordingly, the workload of incumbent judges is increasing. Under such conditions, it is extremely problematic to ensure citizens’ access to justice. If the procedural rights of individuals are not respected, and if an effective, full, comprehensive and open trial is not ensured, trust in the judiciary will not be achieved.”

The announcement by the High Qualification Commission of Judges of Ukraine (HQCJ) mentions a competition for vacant positions in local courts in an even harsher tone: “Catastrophic shortage of judges in local courts.” According to the official statistics of the HQCJ (which is the most accurate one), the number of judges in local courts should be 4,956. At the same time, as of December 9, 2024, there were 1,197 vacant positions (24%). The number of judges in the courts of appeal should be 1,357, whereas the number of vacant positions is 745 (54%!). Even the Supreme Court finds itself in a predicament. There should be 196 judges, but six years after the first competition, there are already 43 vacant positions (22%). This leads to overload of judges, delays in case consideration, lower quality of justice, increased possibility of political pressure and corruption risks, which ultimately leads to loss of trust in the judicial system.

Taking into account the forecasted course of events, on December 11, 2024 , the HQCJ announced a new selection of candidates for the positions of local court judges for an unprecedented number of vacancies: 1,800.

Given the complexity and duration of the procedures for selecting new judges, which are contingent on coordination between the HQCJ, the HCJ and the Presidential Office, it would be too optimistic to expect a quick solution to the problem. In the meantime, there are other technological tools that can simplify and speed up the workflows of judges, their assistants and members of collegial judicial governance bodies. These are artificial intelligence (AI) tools.

What judges fear

Information available in open sources in many languages demonstrates that judges around the world are divided into several groups in terms of their attitude to AI:

1. those who know nothing about AI, except the phrase itself, and are even unreasonably afraid because “it (AI) may someday replace a judge”;
2. those who know nothing about AI but are already actively using it not only at home but also at work;
3. those who have not only heard about it but have also tried to use new trendy generative AI tools such as ChatGPT, Copilot and others;
4. those who actively contribute to the implementation of AI developments in professional life.

We are not going to criticize the representatives of the first group, if only because they do not suspect that they already belong to the second group. At least this can be said about Ukrainian judges, as has ALREADY been officially recorded, even by international organizations.

Surprisingly, this applies to the use of some important technologies with AI elements, which has long been regulated by Ukrainian laws. This is, for example, the Unified State Register of Court Decisions (USRCD), which has been in operation for many years and whose use is provided for by the Law “On Access to Court Decisions” and is carried out by the State Judicial Administration (SJA).

This is also a procedure for depersonalizing information in the text of an electronic copy of a court decision or a judge’s dissenting opinion that cannot be disclosed in accordance with the requirements of the law using specialized software, followed by a selective visual inspection of the results of such masking. This procedure is also carried out by the SJA. It is important to pay special attention to the fact that the USRCD, together with the procedure for depersonalizing court decisions, is included in the database of the Resource Center on Cyberjustice and AI of the European Commission for the Efficiency of Justice of the Council of Europe (CEPEJ).

We believe that this database should also include at least the procedure for the automated distribution of court cases among judges, which has been in place in Ukraine since 2010. There is also an automated distribution of disciplinary complaints among HCJ members and of cases among HQCJ members in qualification assessment, competition and other procedures.

Since 2021, the Supreme Court (SC) has been using the Supreme Court Legal Position Database, which, thanks to the 2023 updates, takes advantage of AI capabilities to search for relevant SC and ECHR case law.

How this innovative development was created:

The key feature of the Database is its full-text search capabilities. In other words, a large text document or a certain text fragment (several paragraphs, a page, several pages) can be uploaded to the Database, which will then select the relevant Supreme Court case law for the available text in this document (according to national legislation, the Supreme Court case law must be taken into account when considering such cases).

The database has an open access and is available to every Internet user, not just judicial assistants and judges themselves. According to some reports, in 2024, the Database will be further improved, and at this juncture, the possibility of using generative AI is being explored. It is necessary to generate a key finding in a particular case from the full text of the Supreme Court’s judgement.

To translate the case law of the European Court of Human Rights, the Supreme Court has also created its own English-to-Ukrainian and Ukrainian-to-English translator, which is trained on official translations of ECHR judgments made by the Ministry of Justice of Ukraine. This translator takes into account the peculiarities of the terminology used by the ECHR and its translation, thus helping to solve, among other things, the problem of translation accuracy.

Therefore, for more than 15 years, Ukraine has been confidently applying AI in legal proceedings and judicial procedures. Moreover, it can be argued that the advent of these systems has been brought about by successful reforms in the area of e-justice.

What about other countries?

Today, it is fair to argue that most AI tools for justice are categorized according to two criteria: those that save working hours and those that reduce the workload of judges.

Chile. In this country, judges are already using Dragon Naturally Speaking (voice-to-text) software, which enables them to make decisions by converting audio into text in real time. It also works with pre-recorded audio files.

The Oficina Judicial Virtual (OJV) platform allows for the electronic filing of lawsuits and written documents anywhere and anytime for all jurisdictions in the country and all levels of jurisdiction.

Experts say that AI is already widely used in the legal sector in Brazil as a common tool for lawyers to monitor cases and conduct legal research on case law.

The Brazilian justice system is deploying a series of AI tools to help reduce the huge backlog of cases.

As of 2019, there were 80 million lawsuits in the Brazilian judicial system with only 18,000 active judges. This means that each judge had to process 4,440 lawsuits per year. In addition, unlike the USRCD of Ukraine, Brazil did not have a centralized database of court decisions available to the public. Consequently, the courts faced significant problems in managing the document flow.

To solve these problems, Brazil’s National Council of Justice allowed the 92 courts it administratively controls to develop their own AI models, which has led to inconsistencies in IT systems across the country. Indeed, courts have created and used at least eight different process management systems.

A survey conducted in 2021 found that initiatives were mainly focused on data structuring and workflow automation to improve the efficiency of court services.

Brazil’s National Council of Justice has created the national platform SINEPSIS, which is used to store and distribute AI developed or recommended by the Brazilian judiciary. The Supreme Federal Court also uses Victor, a chatbot that analyzes and categorizes cases.

Another AI tool already in use in the Brazilian judicial system is designed to facilitate pattern recognition in legal texts (usually in PDF) used by the Supreme Federal Court.

And in a country of 214 million people, more than 1.3 million lawyers are waiting for the Codex system to be deployed, which will collect and store data from electronic court proceedings across the country. This is something that has existed in Ukraine for more than 15 years...

India is very similar to Brazil in terms of the number of pending and backlog cases. However, with 22 official languages, the situation in this country is even more complicated, thus turning the provision of judicial services in the native languages of citizens into a perennial challenge.

The AI Resources Department of the National Informatics Centre (NIC) is actively engaged in modernizing the judicial system. The Neural Machine Translation Service supports the translation of court proceedings into all of India’s 22 languages to improve the operation and accessibility of Indian courts. This tool helps to conduct more than 18,000 district court hearings every day.

The NIC has also created a court transcription tool that supports Indian languages, as well as solutions for judicial search and case law summarization.

The AI tool, known as AI Saransh, is designed to generate summaries of court cases. The tool offers two main types of summaries. An extractive summary highlights key parts of the text from the original content based on predefined indicators without changing the text. In turn, an abstract summary generates new sentences and phrases to convey the most important information, creating summaries that are more natural and closer to human expectations.

Singapore has created a generative AI system that answers questions based on preloaded data. The system is able to inform a litigant how to proceed in accordance with his or her requirements and point to relevant websites and forms.

China, like Ukraine, faces problems with the enforcement of court decisions. To address them, it has introduced systems that track the assets of litigants, ensure compliance with court orders and facilitate the freezing or seizure of assets. For example, the Hangzhou Internet Court uses such an AI system to automatically cross-check national databases. This has significantly improved the efficiency of court enforcement and reduced the number of cases of non-compliance.

The U.S. Court of Appeals for the Ninth Circuit (Florida) utilizes digital reporters to monitor audio recordings of court hearings and produce official written transcripts. They can work remotely and monitor several courtrooms simultaneously.

Attention to the development of justice digitalization and the use of AI tools is paid at the highest state and political level in Europe as well.

Every year since 2020, the European Union has been preparing a Rule of Law Report, which systematically and objectively analyzes changes in the rule of law in all member states. At the same time, the analytical sections on the quality of justice highlight efforts to develop e-justice and, in particular, AI tools. The 2024 Rule of Law Report contains such information not only for each EU member state but also for the four candidate countries.

In the Council of Europe (CoE), the quality of justice falls within the purview of the European Commission for the Efficiency of Justice (CEPEJ), which was established due to the excessive workload of the European Court of Human Rights to reduce this workload by improving the efficiency and quality of justice in member states.

Published on October 16, 2024, the report European Judicial Systems - CEPEJ Evaluation Report – 2024 Evaluation Cycle 2024 (data for 2022) contains reliable data and analysis of the functioning of the judicial systems of 44 European countries (including Ukraine) and two observer states (Israel and Morocco), which allows measuring the efficiency and quality of these systems.

The CEPEJ questionnaire for the Report does not focus specifically on AI tools as such, as is mentioned in the section on Information and Communication Technologies. It does, however, provide in-depth general analysis and samples of some advanced features in various tools, which may include successful attempts at their implementation. These include, for example, speech-to-text functions or “automatic creation of transcripts from audio recordings” of court hearings. For example, the Report states that speech-to-text is already available in 12 CoE member states. Automatic transcription from recordings into text is available in eight states.

The database of the CEPEJ Resource Center on Cyberjustice and Artificial Intelligence classifies the abovementioned and other AI tools into the following types:

1. Search, document review and full-scale research.
2. Online dispute resolution.
3. Predicting the outcome of litigation.
4. Support for decision-making.
5. Anonymization of court decisions.
6. Sorting, distribution and workflow.
7. Recording, transcription and translation.
8. Information and support services.

In lieu of a conclusion

...So many countries (including Ukraine) have now developed various AI tools that are effectively used to optimize workflows in courts. International organizations, of which Ukraine is a member, are also involved in promoting the development of judicial systems in an extremely pragmatic way. And this work is intensifying every month. We can note with some optimism that important strides have been made in Ukraine thanks to the joint work of, for example, the Supreme Court with the EU Project Pravo-Justice in the project mentioned at the beginning of this text.

Attention should also be paid to the draft of the new concept of the Unified Judicial Information and Communication System (UJICS), developed jointly by the State Judicial Administration, the High Council of Justice, the Ministry of Digital Transformation of Ukraine with the political support of the Verkhovna Rada Committee on Legal Policy and USAID Justice for All Activity and EU Project Pravo-Justice.

This means that the use of reliable AI tools (especially national developments) can become both an optimistic component of the national judicial reform strategy and examples of individual successful projects in courts and judicial governance bodies.

The next article will discuss how AI development is defined in government documents at the level of strategies and policies, how it is regulated by laws and regulations, who can become agents of change, and who can prevent the introduction of AI in the work of Ukrainian judges.