Justice lives on a simple and demanding promise: to treat each person as an end and each conflict as a unique, human case, with real-life consequences. The question “what if justice were done by machines?” It is not, therefore, an exercise in science fiction. It is a test of our humanism and the way we currently interpret the constitutionally enshrined slogan “right of access to justice”: how far can automation go without betraying the fundamental guarantees that legitimize the power to judge?
Let’s start with what would seduce politicians and journalists who have never entered a courtroom. Machines don’t get tired, they don’t miss, they don’t postpone due to backlog of work, they don’t have an agenda, they don’t get intimidated (not even by the media). They can research millions of decisions, standardize criteria, reduce territorial asymmetries, cut down time. In a system in which slowness is pointed out by the ignorant horde as the great flaw, the technological promise appears as a way of speeding up justice.
In a system like this, the icing on the cake would be to eliminate the presence of lawyers/defenders of users of Justice, those who, in the opinion of the ex-professor of the horde of ignorance, whose voice thunders as if speaking from a chair, because the stupid is always bold, due to the unconsciousness of his own ignorance.
But, continuing along this line: efficiency is not, in itself, Justice. The jurisdictional decision is not an act of automatism; It is an act of sovereignty practiced with foundation, with contradiction, with responsibility and, above all, as an act of humanity.
An algorithm can suggest patterns; cannot, without risk, replace human judgment. First of all, because the machine learns from the past and the past is not always virtuous: if the data reflects inequalities, the model tends to reproduce them with the statistical serenity of someone who “just” optimizes. Opacity, in turn, threatens the duty to give reasons: if no one can explain why the citizen lost or won, then there is no true syndication of the decision, there is no effective appeal, there is no public trust.
There is also a less discussed but decisive point: institutional empathy. The judge listens, considers and justifies. Justice is not limited to the result; includes the procedure. A “right” decision made by a process perceived as unintelligible or dehumanized is a pyrrhic victory: it is fulfilled, but it is not convincing. Should we, then, reject technology? No. We must put it in the right place.
Automation is valuable in screening, process management, research, detecting inconsistencies, predicting deadlines and reducing repetitive tasks. It can, with clear rules, even increase the transparency and predictability of decisions. But the assessment of evidence, proportionality, the balancing of rights, the dosing of sanctions and everything else that guarantees the effective protection of judicialized law must remain human, precisely because it involves normative choices and not just inferences. The key is the governance of the models to be adopted. They must be auditable, explainable, tested for bias. There must be a record of the versions used, the right of the parties to know when and how the tool influenced the process and, above all, identifiable responsibility. If the machine makes a mistake, who is responsible? If there is no appeal due to the error, the rule of law is compromised and power is outsourced.
In the end, the question gives us one certainty: Justice by machines can mean, in the minds of many, a faster administration of Justice. But justice carried out by machines – automatic, opaque and indifferent – is not a good path.
The desirable future is not the court without people. It is the court in which technology frees up time for the essentials: listening better, deciding better and providing better reasons. Because Justice, to be Justice, must continue to be a human act, under public rules, before equal citizens.
Lawyer and founding partner of ATMJ – Sociedade de Advogados
Leave a Reply