Machines outsmarted lawyers from leading Magic Circle and international firms in a week-long contest.
The Lawyer Challenge, held last week, pitched 112 real life flesh and blood humans against machines fitted with artificial intelligence (AI). Contestants, working unsupervised with all their usual resources, were presented with factual scenarios of PPI mis-selling claims and asked to predict whether or not the Financial Ombudsman would succeed in the claim. The AI team had an accuracy of 86.6%, while human error reduced the real-life lawyers to a 62.3% accuracy scoring.
Ludwig Bull, scientific director at CaseCrunch, the legal AI company which organised the contest, said: ‘These results show that if the question is defined precisely, machines are able to compete with and sometimes outperform human lawyers.’