Oxford Brookes University is developing an app in partnership with corporate technology firm Moorcrofts that can read contracts and identify sections that require human scrutiny.

The two-year project, funded by a government innovation grant, aims to use artificial intelligence (AI) to lighten lawyers’ workload. While AI systems for analysing contracts are increasingly being deployed by major City firms and in-house legal teams at large corporates, the Oxford Brookes team says its product will be aimed at small and medium-sized enterprises. 

The app will employ a statistical technique called topic modelling to identify what a contract is about on the basis of the words it contains. It will then highlight sections that need human attention. Documents provided by Moorcrofts are being used to train the app.

Professor Nigel Crook, head of computing and communication technologies at Oxford Brookes, is overseeing the project. He said: ‘Topic modelling is well-suited to this area as legal vocabulary is typically smaller and better defined.’

The app will also use non-AI technology to help solicitors draft documents. A programme with a ‘built-in editor’, for example, will pull out key information and verify figures.

Chris McCormick, technology and IP paralegal at Moorcrofts, said: ‘There is a lot of work done by lawyers that is not legal work. This app will offload some of the non-legal stuff and check things that often get missed.’

In October a consortium of up to 30 firms will trial a prototype of the app and provide feedback. Firms will pay a fee, as yet undisclosed, to take part in the consortium. When the tool is ready, firms will pay to license the software.

Andrew Katz, partner and head of technology at Moorcrofts, warned that AI could raise ethical issues in the legal profession, giving the example of an automatic negotiation agent.

He said: ‘If technology reaches the stage where it can negotiate between parties, what happens if it learns how one side negotiates to the advantage of the other side?’

The project is one of 40 AI and data analytics projects awarded grants by the government under its industrial strategy.