Will AI become a game changer in litigation? Currently AI is very useful for analysing large volumes of data subject to disclosure or seizure. It can also find case law and literature and do basic drafting. It is not yet good enough in predicting loss and in complex drafting, but can judge work products from external counsel, to be disclosed to the client.
AI can assist both in-house and external lawyers in various ways. First of all it can analyse company data. Examples are due diligence, but in the framework of litigation also disclosure and discovery for countries where this applies. Dutch companies can become subject to the very extensive US discovery, which nowadays is predominantly done by way of an e-discovery. A company that is the subject of e-discovery, which may include thousands of files, will want to know whether the information that is discovered and thereby disclosed to the opposite party contains information that is relevant to the outcome of the proceedings, This is a task that can already by performed by AI.
Actually, as of this year, this has also become more relevant in a Dutch context. Under the new rules of evidence it has become much easier to claim that an opposite party should disclose all information that is relevant to a dispute, see new Articles 194 and 195 Dutch Code of Civil Procedure. In addition, in my intellectual property practice and especially in patent litigation, seizures of evidence are quite common. In a currently ongoing case on my instruction a forensic copy of the servers of the alleged infringer was made. This included more than 350.000 electronic files. AI can be employed to make a selection of the files that are relevant to the infringement case and an analysis of their relevance. The alleged infringer should probably also do that in-house, because it is unlikely that their legal department will be aware of the content and relevance of all of those files.
Obviously, in litigation AI can also be used to find and analyse relevant case law and literature. It may also be used to prepare initial texts for litigation documents, although at the current state of play this cannot replace actual drafting of court submissions in more complex cases. Actually, AI even is not yet sufficiently developed to completely handle less complicated cases. In the UK judgments in employment matters are drafted according to a strict format, which makes it easier to analyse the factors that are decisive for the outcome. Professor Felix Steffek at the University of Cambridge has done an extensive study to evaluate whether AI could predict the outcome of litigation as good as experienced employment lawyers. He had 52,339 judgments available to train his AI tool, of which 11.838 cases were suitable for training. The outcome was a mixed picture: AI was better than human lawyers to predict wins, but worse in predicting loss, at which humans were about twice as good as the most frequently used AI tool. This basically means that AI currently is not yet good enough to predict a company’s exposure.
Drafting court submissions in more complex cases in my view further requires two skills that AI currently does not possess and probably will not possess in the near future: creativity and intuition. Solutions cannot always be reached by applying a syllogism to available data, like AI does. Sometimes you really have to think outside the box. And yes, even in patent litigation you sometimes need intuition to decide what will work, although part of that is probably based on data which AI can oversee, but the human mind cannot.
However, there is one thing that external lawyers could do and that would be very useful for the in-house legal department. Once a court document is drafted a lawyer can ask AI to analyse it and to comment whether there would be arguments, literature and case law, domestic or foreign, that were missed and that could improve the document. The AI report on the final version should then also be shared with the client. Scary for lawyers, but very good for clients.
