The project CaseCruncher Alpha, the evaluation system of the legal queries were rather work with real lawsuits than 100 of the best private lawyers in London. The competition was of an informal character and was intended to assess the effectiveness of a new tool based on artificial intelligence.
AI and lawyers put relatively simple task: based on real data about the controversial payments to assess the likelihood of success when filing a claim for insurance payments. Total was 775 projections on each side, and the AI accuracy was 86.6 %, whereas in humans it is not exceeded, and 66,3 %. For those who want to enjoy legal services, is a significant and obvious difference.
At the end of the competition the BBC raised the issue: will replace Lee in the coming years the AI at the law offices of Junior clerks who are predominantly engaged in the processing of statistical information. The answer is no, but very desirable. Machines are guaranteed superior if the task definition includes a “precise answer”, but not able to renegotiate the conditions of the problem and to think flexibly. But the majority of small employees are performing boring, routine work, and in addition very slow.
CaseCruncher Alpha is the brainchild of three students at Cambridge University, Joseph Marusak, Rebecca Agliolo and Ludwig Bulla. It is noteworthy that all of them are studying at the faculty and do not have deep knowledge in IT. Initially, they designed a chatbot for quick answers to typical questions on claims, but then expanded the algorithm and adapted it for the analysis rather complicated Affairs.
Source — BBC