For a Human-Centered AI

Law as certainty and artificial intelligence: predictive justice

April 27, 2023

Considerations on the possible support of artificial intelligence to the Law, for a justice accessible to all.

  1. Law as “certainty

Ubi societas ibi iusEvery community needs law to live peacefully, as much in existential and social relationships as in economic-operational choices. Only a “certain” legal framework makes it possible to organize the present and plan for the future. What is meant by certainty? How can it be ensured in the modern system? These are just some of the questions that legal experts of all places and times have always tried to answer.

In this sense, the study of the evolution of many legal systems confirms that there is an innate human tendency toward legal certainty. A principle, if not a necessity, whereby every person must be put in a position from which he or she can assess and foresee the legal consequences of his or her conduct.

Unwritten rules of conduct have thus been followed over time by formal and abstract regulatory frameworks capable of governing every situation likely to occur in concrete terms. The a priori prediction of existing rules guarantees – or at least should – legal certainty, ensuring the equality and equal treatment of citizens before the law.

In light of this premise, the question that needs to be asked is: in a context such as the recent one, characterized by tens of thousands of laws regulating the most disparate topics, can citizens’ (usually lacking the ability to understand normative texts) need for certainty be considered satisfied? The answer is harder than it seems. From a purely theoretical point of view, this need can be considered as met since, in fact, there is a legal publication scheme (e.g., Gazzetta Ufficiale or the various Regional Bulletins) that regularly makes the most recent legislative interventions available to the public. However, from a “practical” point of view the matter becomes complicated, since with respect to the aforementioned concept of “certainty,” the rule, in addition to being known, needs to be understood.

Of course, there are figures who help citizens understand the law (lawyers, accountants, consultants …), but their ability to “assess and predict” the legal consequences of a client’s conduct depends on the complexity of the case, is limited to general considerations (no lawyer is able to provide the likelihood, expressed in percentage, of winning or losing a given case), or is influenced by the possibility of making a profit, so that the client convinces him/herself that they should pursue the case anyway.

  1. Predictive justice

The problem is complex and more topical than ever. Before we try to give an answer and explore what is deemed as the solution, we need to start with a simple but extremely topical example.

A low-wage worker, who over the years and with no small amount of difficulty has managed to set aside a limited amount of money, wishes to buy a house. With the help of a real estate agent he finds the solution of his dreams and finally closes a deal worth one hundred thousand euros. However, the first rain causes a flooding in the new house, revealing the existence of a serious drainage problem. So the worker, who had invested all his savings there, decides to turn to a lawyer. At the lawyer’s office, the buyer points out that, during one of the pre-purchase visits, when asked about the possibility that the property might have “water problems,” the agent had replied “not that I know of.” After the seller disappears, the unfortunate buyer considers the possibility of taking action against the real estate agent, but discovers from the lawyer that his case is complicated because the law is more complex than in the past. Even the agent, frightened and in his opinion innocent (since he does not know the state of the drains), turns to a lawyer, and he explains that his case is uncertain, since over time the legal system has been dotted with rules on the professional diligence of agents and on the protections offered to the consumer. In short, the case is uncertain in all respects. Is there any certainty? Yes: both the buyer and the agent, respectively to act and resist in court, will have to pay five thousand euros for the legal expenses (five percent of the value of the property), to which the former must add another five thousand to advance the cost of the repair.

The case that emerges is the following: the worker – now forced, due to lack of liquidity, to take a loan sum – will have to choose between abandoning the dispute and paying the damages of the property out of his own pocket, or using this sum to finance a legal action without guarantees, aware that in the worst case scenario he will have to pay, in addition to the damages of the house, also the onerous legal practice.  In such a situation, in which questions of principle do not come to the fore, would it be desirable that the lawyer already, before the judge, have the means to conduct an ex ante probabilistic assessment, complete with the latest case law, to give the client a clear idea about the likelihood of success of an action? Of course it would, and the solution is called – predictive justice-.

Predictive justice is to be meant as the possibility of predicting the likely outcome of a judgment with the help of specific algorithms (as explained very well in Edoardo Rulli’s article, “Giustizia predittiva,intelligenza artificiale e modelli probabilistici. Chi ha paura degli algoritmi?” (Predictive justice, artificial intelligence and probabilistic models. Who’s Afraid of Algorithms?”). This is a tool to support the legal – and, also, jurisdictional – function capable of analyzing in a short time (shorter than that granted to humans) a huge amount of information, with the aim of predicting the outcome or possible outcomes of a judgment.

  1. Conclusions

If we are wondering about the possibilities of an operational application of predictive justice in Italy, we will first need to find out how it works.  To this end, we will recall one of the most famous predictive justice tools, designed for anyone who intends to exercise claims of any kind in a court of law.

Lex Machina” is predictive software founded in 2009 in California.  It provides particular analysis reports to companies or law firms, aimed at developing programs that will help business improvement or strategy planning for decisions.Through artificial intelligence, lawyers and judges respectively are categorized by “skill” and school of thought.With reference to the former, they are scored: based on parameters such as “expertise” and “overall wins” the chances of success in front of “that” judge compared to another, or, also, in a specific court rather than another.

The fact that technology is critical is beyond question; however, obvious concerns arise as to the actual quality of the assessments.  Indeed, the ability of the lawyer cannot be assessed by taking into account only how many times on that particular type of case that individual lawyer has won or lost, without first considering the context. This is not a sports competition; the diligence and ability of the professional can be maximized even having lost the case since, most of the time, the best (and often the only available) solution is to lose in the “best possible way.” Secondly, the trend to use the network for the creation of a “TripAdvisor” for lawyers will profoundly affect the practice of law, becoming problematic both for professionals, who will be evaluated on the basis of unrealistic parameters; and for clients, since they too may find themselves refusing a case that could have been won instead, and thus be negatively affected by the decisions of the algorithm.

The application of such software in Italy will not be easy, especially if not coordinated with the criteria for the evaluation of a lawyer’s conduct. These criteria, as mentioned above, are not based on wins and losses, but require the professional, in the performance of his or her duties, to adopt diligent behavior (Article 1176 of the Civil Code) such that any expert in the field would follow. On the other hand, if you look at these technologies from the perspective of the client – and not the lawyer – their innovative scope is impressive. Finally the opportunity will be given to all those who cannot afford to entrust their assets to the mere “hope” to make the best choice in the somewhat dubious context of litigation (i.e., the real goal of the legal profession as it was conceived). All by relying on appropriate analysis and statistics capable of providing a spectrum on the probability of winning a given case.

In conclusion, this is a new technology that while raising numerous concerns, at the same time has extreme potential. Removing such concerns will be one of the most stimulating challenges facing today’s jurist and the jurist of the future.



The author/s