Critics had previously said computers could not interpret the complexities of legal cases.
The system was developed by researchers at University College London (UCL), the University of Sheffield and the University of Pennsylvania. The study, led by Dr. Nikolaos Aletras, was published in the PeerJ Computer Science journal today.
The team developed artificial intelligence software capable of detecting patterns in complex decisions. The researchers tasked the computer with weighing up legal evidence and moral questions of right and wrong, enabling it to act as a judge in court cases. The technology is more commonly applied to engagement analysis for films and music but it proved to be adept at reaching legal verdicts.
The algorithm was fed data on 584 humans rights cases covering torture, degrading treatment, fair trials and privacy. The AI then had to make a judicial decision for each case. It agreed with the verdict delivered by the European Court of Human Rights in 79 percent of the cases.
The information extracted from the case topics and the circumstances present across the data were combined to create the overall accuracy figure. To prevent bias and mislearning entering the system, an equal number of violation and non-violation cases were selected. Articles 3, 6 and 8 of the European Convention on Human Rights were used for the study.
The research is the first of its kind. It demonstrates that AI could have far further reaching practical applications than had previously been thought. The technology is becoming increasingly prevalent throughout society. Researchers had previously thought computers would not be able to realise the subtleties and moral concerns of human court cases, however.
“We don’t see AI replacing judges or lawyers, but we think they’d find it useful for rapidly identifying patterns in cases that lead to certain outcomes,” said Aletras. “It could also be a valuable tool for highlighting which cases are most likely to be violations of the European Convention on Human Rights.”
The team attributed the success of the AI to the “realist” approach of the European Court of Human Rights. While developing the technology, the scientists found that judgements made by the Court are highly correlated to non-legal facts, instead of strictly legal arguments. This makes it easier for computers to accurately interpret cases. The discovery supports other studies into the decision-making processes of courts including the U.S. Supreme Court. High-level legal judgements appear to be approached realistically rather than formalistically.
The researchers now intend to further the testing of the system by feeding it more data. They believe it could easily be extended to include the testimonies of witnesses or the notes of lawyers. A better indication of the AI’s accuracy would be afforded if the team had access to the original court applications but these are not publicly available.
“Ideally, we’d test and refine our algorithm using the applications made to the court rather than the published judgements, but without access to that data we rely on the court-published summaries,” said research co-author Dr. Vasileios Lampos.
