Make way for the machines?
Sally Johnson speculates whether artificial intelligence will become more commonly used and accepted in the legal system following recent studies.
A study conducted by researchers at University College London (UCL) and the universities of Sheffield and Pennsylvania has successfully used an artificial intelligence (AI) system to predict verdicts in human rights cases, notoriously full of moral as well as legal conundrums. It had been said no AI would be able to understand the nuances of a legal case but the system was correct (or at least agreed with the judges) 79% of the time. With more data and more refined algorithms, it could do even better.
Lord Justice Briggs will be pleased. His report on the future of the English civil courts has announced funding for the Online Court, a dispute resolution system for smaller claims due to be in place by 2020 which will be almost entirely automatic. The parties will correspond online with a responsive system asking them questions about the case so that it can be prepared for trial without involving lawyers and be put in front of a judge for trial in an efficient manner. Perhaps one day the judge too will be an algorithm. With no waiting lists, fixed fees and a tried and tested system many would welcome it.
Going back to the present day, AI is already in use in civil litigation. With its ability to rapidly identify patterns and to learn from corrections to its results, predictive coding is available as a tool for sorting through scanned or electronic documents in huge disclosure exercises looking for relevant evidence where relevance is a vaguer concept than search terms alone can define. As lawyers come to trust it and use it more, prices are likely to come down and it may soon be an everyday item alongside the sophisticated word search engines we already have
UCL believes its system also has huge potential to help people make decisions on chances of success. If it had access to thousands of decisions a system like that could find any cases with similar circumstances and predict results. The difficulty is that with our common law system, although we rely on historic precedent it is a strength of the system that it allows subtle changes in approach to keep pace with modern times. UCL admits that its systems are not yet capable of understanding nuance. The educated guess the system provides depends on the education and experience the system has had. So perhaps not unlike the lawyer after all, but for the moment it seems the lawyer’s ability to assess not just where the law has been, but also where it is going, will make the difference.
Are we worried this is the end for litigation lawyers? One thing we are sure about is that what comes out will still depend on the quality of what goes in. The machines will still appreciate a well prepared case to crunch the data from. We will just have to get used to predicting how systems think rather than how judges think.
And of course AI still has a lot to learn about negotiation and settlement and weighing up intangible benefits like goodwill gestures or deterrent effect. Or is it that we have a lot to learn from AI and should just let the machines find a logical outcome to disputes while we do something more useful? Whichever way it goes we will be keeping up with the times and using whatever tools we can to sort disputes out as effectively as possible.
About the author
Sally is a Practising Consultant in the Dispute Resolution team at Burnetts.