A more efficient justice through artificial intelligence by Adrien van den Branden
Justice must wake up to the fact that it is falling short of today’s expectations. Citizens around are used to engage in online services that are fast, affordable and of good quality. Courts seem to be sitting at the opposite end. The justice system operates in a paper world. Trials are slow and expensive, and the quality of court decisions is (very) uncertain. This discrepancy, which is growing day by day, could lead to the very end of the idea of a public justice system, citizens preferring to handle their claims over to private (and more efficient) dispute resolution platforms. Worse, disgruntled citizens might no longer want to seek justice at all, for a lack of resources or patience. Hence, we must improve the functioning of courts and tribunals to ensure their survival in the years to come.
How can we build a better justice system?
Absorbed by their daily routine, too few judges or lawyers wonder what a “good” court decision is. Prior to embarking on the mighty task of rebuilding justice, legal professionals must look up and poke at the fundamental question of the meaning of their profession: why do they do their job?
For several years now, I have been reflecting on this question. Drawing on my experience as a lawyer specialised in technology, I came to the conclusion that a court decision is “good” if it is efficient. A good court decision has the best ratio between three factors: quality, cost and speed. A good court decision must be rendered with acceptable quality, within a reasonable time and at an affordable price. Measuring the quality of a court decision is a necessary exercise (and a tricky one as well). The quality of a decision can be measured by criteria such as transparency, impartiality, independence, empathy, conscience, catharsis, reliability, etc. It can also be assessed on two fundamental dimensions (noted by the ancient philosopher Aristotle): the ability to resolve a dispute in compliance with the rule of law (called “retributive justice”) and the ability to resolve a conflict of values, such as the clash between the freedom of expression and the right to human dignity (called “distributive justice”). Finally, court decisions must be assessed on their cost (how much does the trial cost the individual and the community?) and speed (how long will it take for the individual to obtain a final decision?).
Optimising the quality-cost-speed ratio of court decisions
The method for saving justice from its looming privatisation is obvious: optimise the quality-cost-speed ratio of court decisions. Among the means to achieve this, artificial intelligence, i.e. a set of algorithms capable of simulating certain traits of human intelligence, offers interesting prospects. This technology makes it possible to automate certain parts of the ability to judge a case. However, replacing the human judge by a robot judge is not desirable in all cases (far from it). The application of artificial intelligence to justice is acceptable only if court decisions are handled more efficiently than they would through sole human intervention.
If we compare the respective performance of man and machine on all the criteria defining the “good” court decision, robots tend to take more efficient decisions than humans in disputes whose resolution depends on the application of the rule of law (i.e. disputes with a strong link to retributive justice). The outcome of such disputes boils down to the determination of a fine (e.g. for traffic offences), the calculation of an indemnity (e.g. alimonies) or the establishment of payment facilities (e.g. due to indebtedness). Humans tend to render more efficient decisions than robots in disputes that settle a conflict of values (i.e. disputes with a strong link to retributive justice), for example, defamation suits, felonies and the bulk of criminal offences.
Mapping disputes on a retributive justice / distributive justice scale allows us to draw a dividing line between certain disputes that robots would handle more efficiently and other disputes where it would be more efficient for humans preside over. There are a lot of edge cases (e.g. commercial disputes, redundancies for cause, etc.) where the appetite to automate will depend on the political sensitivity of a given population.
The (controllable) risks of a robotic justice system
Transferring automatable disputes to robots is not a risk-free operation. Letting algorithms settle judicial cases may, for example, undermine the right to a fair trial, prompt a loss of court independence or cause widespread disenchantment among the legal professions. However, we can mitigate such risks if appropriate measures are taken to prevent as far as possible adverse consequences from happening. Enshrining a special appeal to a human judge on disputes settled by robots, publishing the underlying logic of judicial algorithms and having the algorithms audited by an independent body are effective measures to protect the fairness of the system.
On the tortuous path of the robotisation of the Belgian justice system, stakeholders will have to take a stand on important issues. The country’s judges, lawyers and politicians must grasp these issues and prepare the courts for the changes to come.
This article was written by Adrien van den Branden is a lawyer at the Brussels bar. He is the co-founder and CEO of Canyon, a tech-enabled law firm providing outstanding legal support to fast-growing companies. Besides, he is a key opinion leader on digital justice and the author of the essay “Les robots à l’assaut de la justice” (Larcier, 2019).
#TheFutureLivesinBrussels by the MIC Brussels. This is a series of curated articles written by experts and partners of the ICT sector and entrepreneurial community. It’s a way for them to communicate their insights on specific topics and share their ideas for a better future in Brussels. If you want to be part of it, get in touch with us!