Predicting justice: what if algorithms entered the courthouse?

A: “I just learned that the entrance to my home, which is rented out, was damaged by the tenants. I want to be reimbursed for the repairs, but can I afford the risk and expense of civil litigation?

B: "Why not rely on an algorithm?”

A: “Algorithm? I would never put my legal rights in the hands of an algorithm!”

B: “Algorithm have neither hands nor hearts, nor preferences or attitudes. They simply perform calculations and provide objective, neutral and reliable responses to your questions – what are the best legal solutions are, and probability that you’ll recover the costs for the repairs. Algorithms can answers such questions.”

A: “Does this means I could go ahead even without a lawyer? It would be reassuring to know I’d receive neutral, objective information. Just one question: who makes the algorithms?”

B: “Well, it’s complicated – I really don’t know…”

A: “So how do we know if the algorithm is correctly designed?”

B: “It’d be based on artificial intelligence…”

A: “But how can we qualify something as intelligent if we do not understand how does it works?”

Is this a conversation out of a work of science fiction? Not at all. Big-data analytics and algorithmic applications are already making their appearance in traditional court settings in many countries, following on the integration and scaling up of information technology (IT) systems already present. One of the leading IT applications is in file management, where it can reduce costs and case speed processing. And yet today, technology seems to be conducive of a much more disruptive and revolutionary type of change.

In France, the possibility of using algorithmic dispute-resolution mechanisms for relatively minor court matters has been included in the justice-reform agenda put forward by the Minister of Justice, Nicole Belloubet. Approximately 2.7 million cases could be affected.

However, the extension of the online-mediation procedure to the cases that have been traditionally handled and adjudicated by first-instance judges – notably small claims and misdemeanours – requires the attention of scholars and practitioners as well as citizens and other stakeholders.

As jurist Antoine Garapon rightly describes it, we are facing a revolution that raises serious questions. This revolution consists into a new method to deal with both the norm-building process and reality construction, as it has been described by Peter Berger and Thomas Luckmann. The digital transformation of how justice is rendered to claimants and litigants touches upon our way of writing the reality, facts and legal arguments traditionally delegated to judges. And what is the minimum level of knowledge and awareness that institutions should ensure that rights-holders have so that they can decide to make a claim or not, or to file a case or not?

Several years ago the Los Angeles Police Department adopted a predictive-policing systemcalled PredPol. While superficially similar to the traditional practice of hot-spot mapping, it uses massive data sets of individual behaviours and neighborhoods, and promises to better direct police efforts to reduce crime. A similar platform, called the Crime Anticipation System (CAS), is being developed in Amsterdam.

Such approaches raise many questions, however. Police work might reduce criminal activities in one place, only to have them occur elsewhere. Or potential criminals could wait until officers leave before taking action. A related approach is risk terrain mapping, where the crime history of a particular region is merged with local behaviours to define crime-prone areas.

Another application that goes beyond simple case management is online dispute resolution, which has been adopted in the United Kingdom as well as in the Netherlands. Many legal scholars and practitioners argue that such mechanisms have a significant and largely underexplored potential.

While IT-driven changes may be challenging, Judge Dory Reiling of the Amsterdam District Court has argued that they have potential for supporting legal and judicial decision-making processes as well as litigation settlements. In a 2017 article on online dispute resolution, she cites several examples:

“Civil Resolution Tribunal is a digital tribunal set up to decide disputes about strata, subsidised housing in British Columbia, Canada. It provides assistance with exploring solutions to a problem, with producing documents, and if necessary, access to a tribunal hearing.”

“Rechtwijzer uit elkaar [legal guidance on separation] helps couples wanting to separate or divorce make a plan for the separation, and with the separation itself. At the end of June 2016, according to information on the site, more than 1,000 couples had started making a separation plan. The tool includes online forms, chat functionality, calculation tools, and the ability to get help from an expert.”

The arrival of such technology can have other impacts as well. For example, once documents associated with civil-litigation procedures are consistently scanned and made available in digital format, large-scale analysis may well be possible as a case of data analytics.

Many legal documents and procedures are already handled by digital tools. However, new laws may push the use of artificial intelligence further.Shutterstock

The knowledge that can be drawn from massive data sets of civil or commercial litigations – and perhaps criminal cases eventually – could also provide information about correlations between situations, behaviours and judicial decisions. For example, patterns of behaviour by litigants, lawyers and judges can be detected and made available to the broader public. Algorithms could also tell us the probability that a legal decision or a judicial adjudication of a case can be set at a particular point, be it costs, fees, sentences, monetary sanctions, and so forth.

But is this everything we need to know, and the story that we want to tell the generations that come after us?

After a long series of reforms oriented toward judicial independence, international bodies such as the United Nations, the European Council and the OECD have promoted an approach that focuses on efficiency rather than exclusively on impartiality.

In this respect, IT-based systems are praised for their perceived potential to reduce costs. Moreover, the possibility to access the justice institutions through online tools – the public portal through which individuals may file a case or follow up the case in which they are litigants or convicted – may be an asset for justice systems that are too dispersed and where citizens may have difficulty traveling to central courthouses.

Figures from the European Council provide an interesting overview as to how and to what extent IT has been infused into court systems.

Information technology can also increase information-access equality. This is what is provided through France’s SAUJ system – Service d’accueil unique du justiciable (“single service for litigants”).

SAUJ is a virtual front desk set up at the level of the first-instance courts where citizens can receive free information, get directions, and access cases in which they’re involved, be it as litigants, victims, witnesses or defendants. Its appeal is that it simplifies access to the justice system.

Introduced in France on an experimental basis in 2014 and followed up in a sample of pilot courts from 2014 and 2016 – Bobigny, Brest, Dunkerque, Privas and Vesoul in France, and Saint-Denis on La Réunion – it has since been extended to the national court system. The use of SAUJ is intended to increase public accountability of the courts, including their management of resources, productivity and organisation.

Yet SAUJ does not tell the entire story on IT’s disruptive impact on the legal system – in both the positive and negative sense.

Pushed further, information technology could provide completely dematerialised tools to facilitate judges’ decisions or better orient the choices of citizens and legal representatives. For example, a judge who must rule on a complex and ground-breaking case may want to be aware of relevant case laws. Data analytics can provide insights, analysing already-adjudicated cases and surfacing those that are similar.

Trends in appeal and revision rates may also be easily detected, and the patterns of past rulings in similar situations could be revealed. Using such algorithms, litigants seeking damages could learn the success of similar cases, and if necessary adjust their strategies or perhaps even decide to not move forward.

While more technology potentially means more insight, will it provide us with not only higher-quality decision, but also more egalitarian ones?

Several questions need to be answered. How would such IT tools be designed, built and monitored? What is the rationale inspiring the creation of such algorithms? And to what extent can citizens be more aware of this rationale than they are of the centuries of tradition that has current methods of adjudication?

What makes the entire difference between the legal code and an algorithm in terms of quality of justice is the possibility for citizens to understand and be responsible in the universe in which they find themselves. Rights-holders are sometimes litigants, sometimes victims, sometimes entrepreneurs, but always citizens.

That is why a digital revolution in the legal world should be observed from the point of view of the citizens, those separate from the algorithms themselves. For years the issue of the opacity and the intelligibility of the judicial machinery has gained the priority in the list of dimensions that a system of quality of justice should consider. In the same vein, if we want to ensure that laws and legal cases are intelligible to citizens and non-legal experts, should we also call for more transparency and responsiveness in the algorithm-making process (in the creation of IT-driven norms)?

A different but equally compelling question is better understanding the point of view of judges and attorneys on the rationale and use of IT tools upon which their daily activities would be based. This will help ensure that we are still ruled by the law rather than falling into the trap of making the rule of law equal to the rule of code.