What if a software could give you the outcome of your trial before it even started? It is every litigant’s dream. “Is it really worth it to embark on a trial whose uncertain outcome is regularly said to be subject to the vagaries of human justice?” How many futile efforts, false hopes and bitter struggles would have been avoided if predictive justice could have told me in advance whether I would win or lose my case.
So be careful, let’s stay on earth. Anyone who imagines that science fiction has become a reality, or that we already live inside a “Minority Report”’s world, in which criminals are arrested even before they have committed any offense, would be grossly mistaken. No, the balance of justice has not yet been changed to a crystal ball or to a virtual fortune teller. Predictive justice is simply a new manifestation of this famous digital transition applied to law.
It is the result of a continuous progress of artificial intelligence, i.e. the ability for a computer program to perform tasks typical of the human intelligence.
The digital revolution of the legal professions is following the movement of the “Legal Technologies”. Launched in North America in the early 2000s, LegalTech has been soaring for the past ten years. The relationship between law and technology is ambivalent. On the one hand, technology reflects the mathematical rigor of certain formal aspects of the law, such as legality or procedure. On the other hand, there are concerns about its inability to deal with more fundamental aspects of justice, such as fairness and the ability to judge. Innovative software to assist in the drafting of legal acts or the search of judicial data allows substantial time savings, while other software aims to robotize the argumentation and decision-making activities hitherto reserved for humans according to just and sound reasons.
Predictive justice is a computer tool used to carry out statistical analyses based on a large amount of data (big data) extracted from case law, i.e. court decisions previously rendered. It shall improve the processing of judicial data transmitted by each court, in particular:
the identification of judgments and rulings that best correspond to the case sought;
the identification of the most effective legal arguments for a given dispute;
the assessment of the chances of obtaining or not recognition of a specific request;
the evaluation of the average amount received by category of compensation;
the assessment of the average duration of the procedure concerned.
Theses programs are based on the exponential growth in the number of digitized court decisions. The more information is collected, the more reliable results and statistics obtained will be. Most efficient ones already gather several million references.
Increasingly fine macro-statistical processing is made possible by increasingly efficient algorithms. The results obtained in record time are translated into graphs and percentage rates that can help legal professionals understand the best strategy to adopt in a given situation. Low chances of success suggest that an alternative method of settlement (conciliation, mediation, etc.) should be preferred to litigation and vice versa. But other variables are also relevant (length of trial, amount of compensation, relevance of arguments, etc.).
From a litigants’ perspective, the main interest of such software is to reduce legal uncertainty, i.e. the uncertainty inherent in any judicial procedure, by providing them with relevant information on similar cases that have already been decided.
From a legal professionals’ perspective, the latest generation of software is supposed to provide a gain in productivity and therefore competitiveness improvements. They facilitate their work on documentary research tasks, allowing them to invest more time on legal argumentation and judicial strategy.
Opportunity ? Nightmare ? or Great Illusion ?
Are we going to leave the task of judging Man to the machine? Will it be endowed with powers of the third kind, with properly human determinants such as consciousness, empathy or intuition?
Fortunately, this not (yet) on the agenda. Because, whatever the qualities and gains that can be expected from such devices, there are obvious limits to the system we should better not to cross since even the most advanced technology would reach an incompetence threshold.
The Courts rule on facts and computer programs do not know how to assess the relevance of evidence and its articulation with the applicable rules. It is one thing to know what the law is, but it is also necessary to establish that the facts correspond to the rules invoked and vice versa.
Justice and law are reflections of an evolutive society. Justice is constantly changing its mind. It even has a name: ‘a reversal of jurisprudence’. The same trial processed 50 years ago could have a very different outcome if it were tried today, even if the law has not changed in the meantime. However, if what has already been judged becomes “The” rule, any evolution and therefore any adaptation of justice to society would become impossible.
The computer answers the question you asked, whether this question was pertinent or not.
Predictive justice does not tell the future but the past, using statistics and other probabilities. It proceeds like these surveys so often use and which are constantly said to be wrong. It is a matter of deciding by looking resolutely to the past while looking to the future in a rearview mirror.
So what can we think about predictive justice? It is an excellent analytical tool, obviously intended for professionals who know how to ask the right questions and put their answers into perspective. It is certainly no more than that. Just as it is not the brush that chooses the painter, but the other way around, it is not “HAL 9000” that will perform our justice. Who doesn’t remember machine's revolutions and other tricks that have inspired cinema for decades? We can’t say we weren’t warned. Once these limits are set, I predict a great future for predictive justice.