What is it?
The use of artificial intelligence (AI) in judicial systems is being explored by judiciaries, prosecution services, and other domain-specific judicial bodies around the world.
Considering the rapid developments in this field, the challenges and opportunities related to harnessing AI in judicial systems and their implications for human rights and the rule of law must form part of the discussions among stakeholders from the judicial ecosystem.
UNESCO is developing a Massive Online Open Courts (MOOC) training course. This is being developed in cooperation with UNESCO’s category-two centre CETIC (Centre of Excellence in Information and Communication Technologies) and the IEEE (Institute of Electrical and Electronics Engineers) — an organisation that promotes itself as the largest technical professional organisation for the advancement of technology) in response to the needs expressed by UNESCO member states, judges, and other stakeholders.
What the project does
Since 2014, UNESCO and its partners have trained more than 17,000 judges and justice officials on freedom of expression, public access to information, and safety of journalists, mainly through MOOC.
These courses reinforce the capacities of judges and other actors of judicial systems on these issues. They provide justice officials with an overview of relevant international and regional legal frameworks as well as new challenges brought by the internet and other digital technologies.
Through a better understanding and knowledge of international and regional standards of freedom of expression and safety of journalists, judges and other members of the judiciary can better align their decisions with international and regional standards, and good practices on these issues.
UNESCO and its partners are developing the programme for capacity building of justice officials concerning the use of AI in courts and by law enforcement, as well as to address the legal implications of AI judicial decisions based on international human rights standards.
Objectives of the online course
Stimulate a participative dialogue with judicial operators on AI-related innovations and promote knowledge of digital innovations in the judicial system
Facilitate knowledge exchange and experience sharing among judicial operators on artificial intelligence, existing norms and standards (hard and soft law) in the field, and its implications for human rights
Highlight existing case studies and best practices that translate ethical principles into practice both in terms of the use of AI in justice systems and in cases involving AI impacting human rights.
Overall, this course will strengthen capacities of judicial operators to address AI-related issues in their domain, ensure they are equipped with the necessary information and knowledge concerning AI-based predictive justice systems, and guarantee that prosecuting services in AI-related cases are aware of the international human rights law as it concerns AI technology and correlated risks.
Commentary
The most common myth about AI is it will lead to decision-making by machine rather than an individual. AI can be described as “allowing a machine to behave in such a way that it would be called intelligent if a human being behaved in such a way.” This is the definition that John McCarthy — who was considered to have invented the term ‘artificial intelligence’ — gave to AI in 1956.
AI is no more than a sophisticated means of analysing and using large quantities of data. AI may also have built-in machine learning processes that will enhance and make the analysis of data more efficient. An example of AI use is in the field of electronic discovery where predictive coding or technology-assisted review (TAR) of large volumes of documents can provide an analysis of a smaller data set comprising documents that may be relevant to an issue in a case.
Within the court context, court cases don’t always require a complex, custom-made approach to decision-making. Many cases can be processed automatically, at least in part. That is why the application of information technology, including artificial intelligence, is not the same for all case types.
Regardless of the subject matter, the work of courts and judges is to process information. Parties bring information to the court, transformations take place during the procedure, and the outcome is also information.
Not all this information processing is complex customisation. Default judgements and statements of inadmissibility are often routinely produced. Many cases require a simple assessment without a hearing, and some cases are settled. Only a limited proportion of the cases that the judiciary must deal with are complex, contradictory cases.
In administrative and civil cases, the way in which cases are handled depends mainly on the complexity of the information and the degree of predictability of the outcome. A relatively large proportion of routine cases have a predictable outcome. In those cases, the court ruling is a document produced in a largely automatic process based on data supplied.
The judgement document provides a basis for enforcement. Here, the court primarily receives digital submissions in which the filing party provides the data digitally, so they don’t have to be re-entered manually. Moreover, if the outcome is predictable, case processing could be partly, or even largely, automated using AI, precisely because the outcome is largely or entirely certain.
The information from UNESCO is of a high-level nature and is rather vague. A survey is being conducted to understand the needs of judges and justice officials that will inform the development of the online course.
The relevance of the course has yet to be determined. Will it educate about what AI is and what it can do? Or will it identify ways in which AI can be deployed in the courts system and in what respect. It may also address issues such as which form of artificial intelligence has already proven itself for these different processes? How can courts and judges work with artificial intelligence according to standards of fair procedure?
It is in the field of data analytics that AI is particularly relevant for the judicial function. The discussion that follows is a very summary of some of the ways in which data analytics can be deployed.
The types of analytics that can be performed with big data are potentially infinite, but can be placed in three categories:
Descriptive analytics: This is the traditional form of analytics. Often it involves the application of filters before applying simple mathematical operations to condense the vast amounts of information into more easily understood information. This allows us to gain an insight into what has happened. This can range from information as basic as the number of views a YouTube video attracts to more detailed information such as a company’s financial reports.
Predictive analytics: It essentially involves the use of existing data to generate a model, which can then be extrapolated to predict likely outcomes in a given set of parameters. In other words, it uses data to predict data that is lacking. This allows us to forecast what could happen.
Prescriptive analytics: This is an evolution of predictive analytics. Instead of predicting a likely outcome based on one set of parameters, it involves multiple predictive models running in parallel based on different sets of parameters. In a decision-making context, this allows the decision-maker to gain an insight as to what should be done to obtain the best possible outcome.
Machine learning
A subset of artificial intelligence, machine learning uses algorithms to parse data and learn from it without being explicitly programmed. The term has come to be synonymous with predictive analytics, but a distinction is drawn here: predictive/prescriptive analytics shows how big data can be used, while machine learning provides the technique to allow such use. Within machine learning is deep learning. This is where the machine is fed an algorithm together with great amounts of information. The machine then attributes different characteristics and draws its own connections to understand the data. In other words, the machine teaches itself.
A couple of examples of AI applications follow:
CARA — a tool developed by Casetext, powered by machine learning — completely sidesteps the search process in legal research. It analyses a legal document that is fed to it and generates a list of authorities that are relevant but have not been cited. It obviates the need for parties to key in search terms in respect to the issues raised and provides a powerful way of detecting weaknesses and omissions in legal documents. But it has its weaknesses. Because it relies on correlations between cases, such as how often they are cited together, documents citing only new cases may not generate results that are as effective.
Another real-world application of data extraction is Kira, which is being used by Deloitte and is the subject of agreements with DLA Piper and Clifford Chance. It uses machine learning to search large sets of unstructured data to identify concepts and clauses, which make it easier for a user to identify issues and trends across different documents. This has huge potential for in-house knowledge management.
Predictive analytics would have a use in the sentencing process, especially in analysing sentencing data contained in cases and developing possible outcomes or a range of outcomes based on the available data. Care would have to be employed to ensure that the decision is that of the judge and not that of the machine.
Conclusion
Involvement in this process must have a useful and practical outcome for the New Zealand judiciary and their support people. Given the push back against technological tools and the expanded use of technology in the court process, there may well be little appetite for involvement in this programme.
There is not a lot of point in employing judicial resources in learning about the utility of processes that are not going to be used in the foreseeable future. That said my own view is that this should be pursued.
The vigour with which Judge Noel Cocurullo has pursued the development of centralised working in the Family Court and the continuing online hearing process using Teams and Sharepoint demonstrates there is an appetite for such developments and for technological solutions. The UNESCO project may keep that flame alive.
At this stage I do not have enough information to suggest that the IJS be involved. As you can see, the project is still in the developmental stage. But when some crystallisation of what is to happen is in place, consideration could be given to enlisting their support.
Recommendations
Continue to maintain a watching brief on the project and its continued development
Explore via JANZ and its contacts whether other jurisdictions are or are going to be involved, and then start a line of communication as to their views on developments
Ascertain what interest, if any, there might be on behalf of the New Zealand judiciary in learning about AI use as part of the judicial function or in courts administration. This could be done via a JANZ enquiry.
Mandate
JANZ authorises the appointment of a liaison judge to interact directly with UNESCO on this project and provide a watching brief upon developments with a report by 1 June 2021 or earlier on another jurisdiction’s activity in AI. This comes with a view to establish a standing committee who will promote communication and exchange of ideas on AI and the judicial role.