Artificial intelligence - quantifying legal risk

Changed on 28/10/2019
Divorce. Car accidents. Unfair competition. Wrongful dismissal. Many disputes result in one party being forced to pay a sum of money. An indicative scale is sometimes used to calculate the totals to be paid, something magistrates draw upon when it comes to reaching a ruling. In practice, however, there is often a discrepancy between this figure and the total that is actually paid, reflecting the particularities of each individual case that judges must take into account. As a result, it can be difficult to estimate the total amount of compensation in advance.


L'intelligence artificielle développée par Case Law Analytics vient réduire cette incertitude. Comment ? « En étudiant la jurisprudence, explique le mathématicien Jacques Lévy-Véhel, cofondateur de l'entreprise et ancien directeur de recherche à Inria. Notre outil va explorer les jugements dans les différentes juridictions de France, observer la réalité de terrain et modéliser la décision. Il va ainsi être capable de donner un éventail de probabilités quant au montant de l'indemnité qui serait accordée sur un nouveau dossier. Notre IA va dire, par exemple, que sur tel contentieux, il y a 70% de probabilités que la cour d'appel de Rennes attribue une somme de tant, 20% de probabilités qu'elle accorde une somme de tant, etc. »

The artificial intelligence developed by Case Law Analytics aims to reduce this level of uncertainty. But how does it work? “By studying case law ”, explains the mathematician Jacques Lévy-Véhel , co-founder of the company and former director of research at Inria. “Our tool aims to explore the judgements reached in different jurisdictions across France, to observe the reality of the situation in the field and to model decision-making. This will enable it to deliver a range of probabilities with regard to the total amount of compensation to be paid within the context of a new case. Our AI will state, for example, that for a given dispute, there is a 70% chance that the Rennes Court of Appeal will award a certain sum, a 20% chance that it will award another sum, and so on and so forth. ”

Decision support

Launched in September 2017, the company quickly found its market. “We work for the legal departments of major organisations, such as the SNCF, or insurance companies such as AXA Legal Protection. For these companies, our AI provides an additional decision support tool. They use it to enhance their risk provision. Our solution is also used by law firms in order to refine their adjustment operations and to perform simulations. In some cases, the tool gives them figures that they can use to convince clients that it would be in their interest to reach a settlement or to compromise with the other party, given the likelihood of the total amount of compensation falling short of their expectations. Letting clients see this risk quantification with their own eyes puts them in a position of being able to make a more informed decision. ”

With premises in both Nantes and in Paris, Case Law Analytics is part of a group of eight start-ups housed within the AI Factory , an incubator for artificial intelligence launched by Microsoft within Station F, the digital campus set up by Xavier Niel, the founder of Free (a French internet service provider and mobile operator). “In addition to the premises, Microsoft have provided us with materials, their IT infrastructure and, most important of all, their address book, putting us in touch with some really first-rate contacts. ”
 The company is currently working on other risk quantification products for use in a legal context as they prepare to expand across Europe. “Our model is capable of functioning in other legal systems, not just in France. ” Case Law Analytics already has eight people on the payroll.