E-justice: salvation or a new form of bias?

With the rapid development of machine learning and artificial intelligence technologies, supporters of robotization of any processes also thought about the improvement of the justice system. The outdated and often excessive regulation, as well as the protracted period of trial by judges in Russia, to which only a share of confidence remained — all points to the need for future reforms.

sourse image

Of course, we are talking not only about legislative changes, technology that allows you to analyze a huge amount of data and tell a person the probability of the applicability of certain rules in a particular case — is simply necessary. The primary task is to transfer all "indisputable" cases, when the mass of court orders on debts, Bank claims and other claims takes time from truly fateful decisions.

It is no secret that even experienced judges with a well-trained staff of assistants spend most of their time on a selection of legal formulations. In these conditions, they often have to work "on the basket", because even on the motivating part of the decision already taken sometimes given weeks. The consequences of attempts to "simplify the process" are well known, the template phrases and superficial consideration of cases, causing serious damage to the interests of citizens and the reputation of the entire judicial system.

The current training of many judges in Russia is sad, the backbone of the newcomers is the Executive staff, who have long accepted the illusory independence of decision — making. I do not dare to denounce them as "distance education", but sources are alarming, most of the judges came from the same apparatus and law enforcement agencies. In the system every year becoming less the real "foreign experience", however, perhaps the goal is not the recruitment of experienced lawyers, and "obedient artists"?

Russian "innovations" stopped at the level of banal audio streaming and speech recognition, allowing at least a little to reduce the work of assistants. As before, the bulk of the funding goes to the repair of buildings, improving the living conditions of judges, equipping office equipment and other material component. The judicial Department does not hesitate to use an extremely opaque it-procurement market and any developments about which the public is "not supposed" are submitted under the sauce of "artificial intelligence".

Technologies that use machine learning certainly have great potential in legal matters. The same Caselook system helps me (a person with little practice in recent years) to see what decisions were made by the "appointed" judge, and to predict more effective arguments in the process of interaction with him. But can they in the near future for something more, in addition to summarizing information, identifying patterns and drawing the model claims?

Experts who have demonstrated an algorithm capable of predicting the decisions of the European court of human rights found that judges almost always pay attention to the very essence of the problem (the moral component), rather than strict adherence to the facts and the letter of the law. Does this mean that the use of an automated system without Titanic recycling legislation can increase the chances of an unfair result, rather Yes than no.

"Emotional" artificial intelligence also does not stand still, the algorithm for videotaping lawsuits in 14 of 15 cases is able to determine the characteristic human emotions lie. It is worth noting that the program does not analyze the speech itself, but takes into account only facial expressions and facial expressions. Well, a possible Arsenal for the trials of the future, is impressive.

However, the algorithm for predicting recidivism COMPAS, used in practice in the United States for many years, constantly causes more and more litigation. About bias of an assessment of the facts, erroneous data (provided by law enforcement agencies), too generalized conclusions, a question of sex and race and many other things.

I am not an alarmist, but is it worth blindly trusting such questions "machine" when the actions of people are less and less amenable to logical evaluation. To what extent is strict adherence to the rules (often ambivalent), without taking into account all circumstances and conduct? I think modern automation has its limits, but a good "shake-up" of conservative jurisprudence will not hurt.

It is absurd, when already in many corporations claims are prepared bots, and the effectiveness of arguments on a particular judge is estimated by third-party commercial online services. The judges in Russia require more papers with stamps, assistants are asked to bring templates for ease of copying decisions, and the postal receipts are a counterargument in the debate for evidence.

Whether the choice of the society, between indefatigable and impartial algorithm in a struggle with a miscarriage of justice, corruption and vicious factors?

It is about the transition of justice of human passions to a kind of uniform mathematically accurate legality. Undoubtedly, such changes will not only increase speed and efficiency, but also contribute to greater transparency, and hence our confidence in the judicial system. But where to draw the line, a dividing line between a boring routine and true justice?

And finally, more of a rhetorical question, if humanity still has not managed to reach a fair and impartial trial, why do we think that we will be able to teach artificial intelligence? :)