Tom Salmon
It is a cold Wednesday morning. Your fingers tingle in anticipation. The day of your court case is here. You sit on the couch in your living room. You open your laptop and login to the NSW court’s website. A virtual courtroom suddenly appears! You see the slightly distorted faces of your lawyers, adversaries and public observers through their webcams. You also see an odd, almost-human looking judge. This is no ordinary judge. This judge is a computer.
Integrating artificial intelligence (AI) judges that simulate human intelligence into the legal system could save resources and prevent delays. However, from a human rights perspective, systems of AI judging may infringe upon the right to a public hearing if they prevent the observation of legal proceedings by members of the public.
AI and the legal system
There is an increasing use of AI in legal processes throughout the world. Some law firms use AI to assist in the ‘e-discovery’ of documentary evidence in civil cases. In some instances, AI even provides legal services. DoNotPay professes to be the ‘world’s first robot lawyer’ and assists clients in tasks such as challenging parking tickets and filing small claims matters.
It is unclear whether any countries have implemented AI judges. The Estonian Ministry of Justice recently denied reports that Estonia was developing an AI judge for small-claims matters worth less than €7000. Nevertheless, the application of AI in judging appears promising. In 2016, researchers developed an AI program which predicted case outcomes of the European Court of Human Rights with 79% accuracy. In 2017, another research team developed a similar model that could determine whether the Supreme Court of the United States of America would follow or reverse the ‘status quo judgment’ in a case with 70.2% accuracy.
What is an AI judge?
Some academics consider that the integration of AI into judging should be limited to tools assisting human judges. In contrast, a truly autonomous AI judge would perform all functions of a human judge. These include determining the facts of a case, determining the relevant law and applying the law to those facts.
AI judges would draw upon data from previous cases and other sources of law to analyse a case and determine its outcome. AI judges may employ machine learning to learn through experience and make decisions according to up-to-date data.
Models of AI judging
AI judging has been widely discussed in academic literature. However, analyses are often limited to examining how AI judges use data to produce an outcome. While these discussions raise many valid concerns, such as the potential for AI judges to discriminate, they rarely consider how autonomous AI judges would be integrated into trial and hearing processes.
A few potential models for integrating AI into judging come to mind. For example, an AI judge could make decisions purely on documentary evidence submitted to an online portal. Another possibility, as explored at the beginning of this article, is that litigants could appear online through a video-conferencing platform before members of the public and a holographic judge who gives reasons for decisions. This would closely mimic the virtual hearings undertaken in Australia during the COVID-19 pandemic.
The right to a public hearing and AI judges
The right to a public hearing, often viewed as part of the broader right to a fair trial, is embodied in many human rights treaties. For example, Article 14 of the International Covenant on Civil and Political Rights (ICCPR) states:
In the determination of any criminal charge against [a person], or of his rights and obligations in a suit at law, everyone shall be entitled to a fair and public hearing by a competent, independent and impartial tribunal established by law.
The Human Rights Committee of the United Nations provides further details on the application of Article 14 in General Comment 32 to the ICCPR. It provides that trials ‘must in principle be conducted orally and publicly’. It also states that the ‘[c]ourt must make information regarding the time and venue of the oral hearings available’, ‘provide for adequate facilities for the attendance of interested members of the public’ and make public ‘the judgment [of a case], including the essential findings, evidence and legal reasoning.’ There are limited exceptions.
As can be illustrated with reference to the two models raised in this article, systems of AI judging may erode or preserve the right to a public hearing. The right is infringed where parties submit evidence in documentary form to an online portal for an invisible, AI judge to decide the case without public observation. However, the right is effectively upheld in systems where, similarly to a human judge, the public can observe AI judges administering court proceedings and giving reasons for decisions.
Systems of AI judging should preserve public hearings
It is paramount that systems of AI judging preserve the right to a public hearing. The right to a public hearing promotes open justice and transparency, upholding both ‘the interest of the individual and of society at large’. As Justice Gibbs of the High Court of Australia stated in Russell v Russell, public observation of court proceedings subjects judges to ‘public and professional scrutiny and criticism, without which abuses may flourish’. In the AI judging context, public hearings would ensure governments and those responsible for coding AI judges are accountable for the conduct of AI judges. This may result in, for example, attention being paid to ensuring processes such as fact-finding are fair and accurate. The openness of proceedings would also promote public confidence in AI judges which may otherwise be subject to scepticism.
Implications
If autonomous AI judges are to be implemented around the world, governments must be careful to preserve the right to a public hearing. This is only achieved if AI judges conduct court proceedings which are observable by members of the public and give public reasons for decisions. Developments in AI technology are necessary to make this possible, such as the creation of ‘explainable AI’ which allows people to understand an AI’s reasoning for a decision. Otherwise, legal systems risk compromising people’s right to a public hearing to achieve sheer efficiency.
Tom Salmon was an intern with the Australian Journal of Human Rights for Term 1 2022.