More than 50 solicitors, barristers and in-house counsel have volunteered to outsmart predictive software based on artificial intelligence in a ground-breaking lawyer v machine challenge.
Lawyers will have a week to predict whether real PPI complaints were upheld or rejected by the Financial Ombudsman, before CaseCrunch, the new name for what was Elexirr, predicts the outcomes in its own way.
Ludwig Bull, managing director of CaseCrunch, told Legal Futures: “We see this as an experiment. It’s not so much about winning or losing as making a lasting contribution to the field.”
Mr Bull said his team chose a “very narrow” area of commercial law because almost all the lawyers signing up were from commercial firms and it had to be an area where non-specialists could acquire the knowledge they needed very quickly.
He said the challenge would last for a week, beginning on 20 October, and lawyers could make as many predictions as they had time for. He estimated that lawyers might take around two hours, making 10 predictions.
Mr Bull said there was no limit on the number of lawyers who could sign up for the challenge, although the minimum of around 40 had now been passed, or on the number of outcomes they could predict.
Mr Bull said he was “so confident” in CaseCrunch’s technology that he was not concerned by lawyers using their own technology, taking a lot of time on their predictions or even collaborating. He said one City law firm had asked to enter a team for the event.
“We did consider doing this in a closed environment, but in the real world we’re up against lawyers with all their capabilities.
“There is no other technology like ours in the world at the moment. None of it can solve problems the way we can.”
Mr Bull said predictive analytics pioneers Premonition would act as technology judge for the competition, examining the code CaseCrunch was using and the predictions by lawyers.
Just like the machine, the lawyers will work on randomised complaints – the main difference being that the machine can make thousands of predictions in a minute.
A legal judge will also be appointed to ensure that the task is one that could be tackled by lawyers and that the machine does not have an unfair advantage.
Mr Bull said: “We’re doing it to win, but what we care most about is making good progress on legal AI. It’s an area that has been so neglected and there has been so little fundamental progress beyond the hype.
“In legal AI, for some reason, the benchmarks you see elsewhere don’t exist. We want to get the ball rolling and, if we lose, at least we know where the benchmark is and we can improve.”
To sign up to the challenge, click here.