Dennis Byron | Beyond digital transformation: Justice in the age of AI – Part I

We can confidently expect that it is only a matter of time before AI becomes more prevalent in Caribbean courts and justice systems.

Caribbean courts have increasingly embraced the use of technological solutions in delivering justice. Virtual hearings, digital case management software, automated recording of proceedings, and online payments have all been markers of the evolution of justice delivery. And now, as we enter into the age of AI I believe that we will witness another stage in this evolutionary process., not merely transforming how we work but also challenging how we view justice delivery and court systems.

It is reported that in 2017, China established its first Internet court, in the city of Hangzhou, to resolve cases virtually using non-human judges, powered by artificial intelligence (AI) and operating 24 hours a day, seven days a week. in 2023 it was reported that 3.1 million legal activities were completed. Meanwhile it is reported that in Beijing, the average duration of a case is 40 days; the average dispositive hearing lasts 37 minutes; and 98 per cent of the rulings have been accepted without appeal.

Traditional legal training in the Common Law often suggests that too much speed is like too much light, which blinds, and would question whether these results could co- exist with fair trials and all the generally accepted core judicial values: equality before the law, fairness, impartiality, independence of decision-making, competence, integrity, transparency, accessibility, timeliness, and certainty.

We must consider both the opportunities and the risks AI presents. The challenge before us is clear: how do we leverage the immense potential of AI while avoiding potential pitfalls that may jeopardise public trust and confidence?

SURVIVING DISRUPTION

The potential for bias in AI algorithms is a significant challenge to fairness and impartiality.. A 2017 study demonstrated that AI can be programmed to have racist or sexist biases. An example in the US is COMPAS – the Correctional Offender Management Profiling for Alternative Sanctions – an AI-driven tool used in US courts to predict the likelihood of a defendant reoffending. But an investigation revealed that COMPAS exhibited racial biases, disproportionately assigning higher risk scores to Black defendants compared to White defendants with similar profiles. These concerns were judicially discussed in State v Loomis (2016) a Wisconsin Supreme Court case, challenging the use of COMPAS in sentencing. The court upheld the sentence but acknowledged the tool’s limitations, mandating that any use of COMPAS in sentencing must be accompanied by warnings about its potential biases and lack of transparency.

Another pitfall is the use of unverified AI content. In Mata v Avianca, a personal injury case against Avianca Airlines, heard in a New York District Court in 2023, an attorney submitted a brief that cited non-existent case precedents generated by ChatGPT. The court dismissed the motion and imposed a fine of $5,000 highlighting the dangers of unverified AI-generated content in legal proceedings.

The technology is still developing, and some judges have highlighted the need for vigilance and for ensuring that attention should be paid to the principles of due process, necessary to ensure fair trials. Cybercheck is a developing AI investigative tool used by law enforcement agencies to geolocate individuals which helps to prosecute and convict persons charged with serious crimes. It becoming widely used, in nearly 8,000 cases. But some defence attorneys have criticised its use and it has been reported that a judge in New York and one in Ohio barred the introduction of Cybercheck evidence because they considered that its reliability had not been sufficiently well established . On the other side of the spectrum, a Judge in Washington excluded AI-enhanced video evidence in a criminal case that the defendant had tried to introduce to prove his innocence, stating that the software lacked peer review and independent reliability testing, and that its algorithms were “opaque and proprietary.”

I have referenced cases in the US where the standards for acceptance of this type of expert methodology was set by the US Supreme Court in 1993 in the case of Daubert, Merrell Dow Pharmaceuticals before AI was as developed as it is today. The factors for acceptance included the extent of testing and whether it was subject to peer review or had a known or potential error rate and was accepted by the scientific community .

PRESSING CONCERN

Another pressing concern is the potential loss of the empathy and contextual understanding possessed by human judges. In the Caribbean context, where cultural and social factors deeply influence disputes, this loss of human judgement could have far-reaching consequences.

Data privacy risks further complicate the adoption of AI in justice systems. Newsday Trinidad reported that 2022, the Caribbean experienced more than 144 million cyberattack attempts, with ransomware being the most common breach .

Finally, we must confront the challenge of institutional resistance to change. The adoption of AI demands a paradigm shift that many justice systems, are hesitant to embrace.

Together, these disruptions present a formidable challenge, but they also offer an opportunity to reimagine justice delivery and navigate these challenges. The path forward must be built on education, collaboration, experimentation, and ethical foresight.

NAVIGATING THE CHANGE

The cornerstone of navigating this transformation is education and capacity building for judges, lawyers, and court administrators. And certifications that could help professionals differentiate between valid AI-assisted tools and those that pose risks to justice.

Collaboration is equally critical. Without reinventing the wheel, and drawing inspiration from existing initiatives such as the European Union’s ethical guidelines for AI, the Caribbean can develop localised policies that prioritise fairness, accountability, and transparency. These principles, embedded within the region’s legal framework, can guide AI’s responsible use in justice delivery.

Ethics must underpin every aspect of this journey. AI systems used in justice delivery must be built with fairness, accountability, and transparency as non-negotiable principles. Mechanisms for oversight, must be established to hold AI systems accountable for their outputs.

Navigating this change requires a clear vision for how AI can enhance justice delivery without compromising its foundational values. The challenges may be significant, but so too is the opportunity to shape a justice system that not only survives disruption but thrives in the face of it.

Sir Dennis Byron is the former president of the Caribbean Court of Justice, and former president of the International Criminal Tribunal for Rwanda. This two-part article is from his keynote address delivered at Connected Caribbean Summit in Miami in December 2024. Send feedback to columns@gleanerjm.com