Students say ‘chatbot’ can predict case result better than lawyers
Leading solicitors dispute reliability of program created by undergraduates
A GROUP of Cambridge law students have become embroiled in a dispute with some of Britain’s leading legal minds by creating a “chatbot” that they claim can predict the outcome of a case better than a lawyer.
They say that the “bot”, a computer chat program that analyses users’ legal problems, could be used to match people with law firms to fight their case.
However, their invention has brought them into conflict with leading solicitors who say they are failing to take account of the “nu- ances” of the law. David Greene, a senior partner at Edwin Coe, the London law firm, claimed the students didn’t have “any regard to how legal service is delivered” after they suggested their “chatbot”, Elexirr, could outsmart some of Britain’s best legal minds.
The bot, a computer chat program that can be used through Facebook Messenger, predicts the likelihood of the user winning their claim, by asking a series of questions about their legal problem.
Nadia Abdul, Rebecca Agliolo, Ludwig Bull and Jozef Maruscak, who are all law students at the University of Cambridge, plan to stage a “man versus machine” challenge in October. Both sides will be given legal problems, with the aim of predicting the outcome.
However, Mr Greene said the students have missed the reason why lawyers are hired in the first place. He also claimed they’d missed the “nuance[s]” of law.
“When people come to us [lawyers], they don’t just want to be told, ‘well, you might lose this case’. What they want to be told is how they’re going to win,” he said. “That’s where the machine fails.
“They’re young students who might consider that law works in a mechanical way. Well the answer is it doesn’t.”
Mr Greene also questioned whether the chatbot could accurately predict a judge’s decision. “There’s an old saying… that the decision of the judge might be dependent on whether he’s had bacon for breakfast or not,” Greene quipped.
The students say that the chatbot, which took around 10 weeks to build, can correctly predict case outcomes 71 per cent of the time.
However, according to Mr Greene, legal advice will be given “person to person” for the foreseeable future.