Judges in the sights of robots

Artificial intelligence improves the work of the judges to such an extent that, one day, it may even replace them, analyzes lawyer Caio Cesar Rocha.

The year is 2022. In the midst of the presidential dispute, Brazilian researchers develop a computer program to take the evidence on the 45 sentences of ex-judge Sergio Moro at Lava Jato. The machine, equipped with artificial intelligence, is capable of calculating the degree of isonomy of decisions and pointing out those that deviate from the standard, considered more biased. So, what do you think the result would be?

The story, of course, is fictional. But, believe me, it is much more plausible than a lot of fake news. Today, there is already technology and scientific knowledge to conduct this type of experiment. Similar tests were carried out three years ago by a group of computer scientists at University College London in England. They created a robot judge capable of predicting the sentences of 584 European Court of Human Rights cases.

The supercomputer learned the legal vocabulary and was provided with a database of the court’s previous deliberations. He also went on to codify the arguments of lawyers, prosecutors and judges. Finally, he began to correlate the facts described in the case file, the circumstances in which they occurred and the current legislation. Result: 79% of the decisions were correct.

This is not an isolated feat. In previous experience, scientists at the Chicago-Kent College of Law, of the Illinois Institute of Technology, created software capable of predicting, with a 70% accuracy rate, how the United States Supreme Court decisions would go. But how is that possible?

Basically, artificial super-brains have developed an unparalleled ability to understand the patterns that govern human judges’ decision-making processes. Deviations from the standard, therefore, could indicate the bias of the judges in their convictions. By bias, in fact, we understand: superimpose the application of the law in a case by, for example, political or ideological activism, religious belief, prejudice, conflict of interest, relationship with one of the parties or some other misconduct.

To continue our fictional story – but, say, based on real events – Justice Minister Sergio Moro, caught in questionable conversations with Lava Jato prosecutors, could have his exemption challenged by the algorithms. His sentences would then be corrected. He wouldn’t be alone in this. Digital scrutiny would apply to all members of the judiciary. Years later, they would end up losing their jobs to robots. Have you thought?

Robot Judge vs. Human Judge

According to researchers in English and American studies, robot judges cannot yet replace a human judge. Legal codes, unlike computer codes, are subject to interpretations that vary depending on the facts involved in a court case.

You have certainly heard the expression “each head, one sentence”. Yeah. This is especially true for cases with greater complexity, in which decisions are differentiated by the sensitivity, understanding and knowledge of those who judge.

Law is not an exact science. It is a knowledge in constant evolution, which accompanies social changes and the way we interpret moral and ethical issues. It requires a certain flexibility of reasoning that machines do not yet measure. Non-standard deliberations, therefore, could not only mean misconduct, but a more innovative and passable type of legal interpretation.

What weighs in their favor, however, is the possibility of eliminating human bias and bias. This search is old. At the end of the 18th century, so-called codicists believed that it was possible to rationalize laws in such a way that legal systems would bring hypotheses and previous solutions capable of resolving all types of conflict. However, the complexity of human relations ended up managing to crush this utopia.

Now, disguised as technological innovation, the thesis seems to gain strength again. Computers are already being used by the Judiciary from different places. This is the case in Estonia. The European country is implementing a computerized system to analyze legal disputes involving less than 7 thousand euros. The parties to the dispute send the documents and the computer decides who is right. The sentence can be reviewed by a magistrate.

In the United States, economists at the National Bureau of Economic Research have developed an algorithm capable of estimating the potential of criminals to commit new crimes while awaiting their probation. The resource proved to be effective. It reduced recidivism by 25% and did not increase the prison population. In practice, the system proved to be more efficient for assessing risks than human judges.

Something in this sense would be very useful in Brazil. There are currently 714,000 inmates in the country. One third of this population is made up of people who have not yet been convicted, but are awaiting trial. A significant portion of these people could wait in freedom without posing risks to society and, on the other hand, without contacting the most dangerous criminals in jail. According to scholars, this would reduce the number of pre-trial detainees by up to 40%. It is the equivalent of 95 thousand people.

But even robots can be biased. An example of this also comes from the United States. Another system used in different American states to assess the risks of recidivism is accused of perpetuating old prejudices in the judicial system. Unlike the program of the National Bureau of Economic Research, it is a little transparent algorithm, without external control.

Brazilian robots

In Brazil, the National Council of Justice (CNJ) announced in February the creation of an innovation laboratory and an artificial intelligence center. The idea is to develop programs that facilitate the day-to-day life of Justice and speed up the process. The Federal Supreme Court (STF) and the Superior Court of Justice (STJ) have similar initiatives. And some state courts, such as those of Minas Gerais, Paraná, Pernambuco, Rio Grande do Norte and Rondônia, are also advancing.

In such cases, in general, robots are used to make life easier for judges and to decongest courts. This is the case of Radar, created by the Minas Gerais Court of Justice. The system gathers 5.5 million cases and allows magistrates to check repetitive cases in the counties, group them and judge them jointly based on a standardized action.

The next step would be to eliminate the need for decisions to come only from the magistrate. As in the Estonian example, artificial intelligence could be used in Brazilian courts to resolve minor and less complex disputes.

In areas where there is a significant increase in judicialization, such as health, where there has been an increase of 130% in the number of demands in ten years, the algorithms would be in charge of defining sentences on a large scale, quickly. Everything, of course, with the supervision of the judges.

There are many paths to be followed. In most cases, these are advantageous options. Robots reduce the time it takes to process cases, assist the work of judges, can replace them in simpler matters, make justice more predictable, reduce costs and correct human errors. All of this with the benefit of not exchanging messages with the parties involved in the process, nor claiming a position in the Supreme Court.

Source: Migalhas website: https://m.migalhas.com.br/depeso/307179/caio-cesar-rocha-juizes-na-mira-dos-robos?fbclid=IwAR3mVK6yBtY9j4KwXU4WAZHoHTC5Vt7h4YS2KOjaIRCMrpJKeWI8y