First Leverhulme Early Career Fellowship for School of Law explores sustainable methods of legal AI

The Leverhulme Early Career Fellowship is offered to early career researchers to undertake a significant piece of publishable work. This project investigates whether the natural human language of law can be successfully translated to machine-readable form with potential AI (Artificial Intelligence) capabilities.

Machine-assisted solutions in AI for the legal industry are increasingly successful in automating simple, repetitive and time-consuming legal tasks. The design of these task-specific solutions, however, results in largely incompatible technological fragments.

To rigorously assess the capacity of AI to process and engage with the intricacy of legal language and concepts, Ribary’s three-year project will be carried out in the laboratory conditions of a closed and theoretically complex system of ancient law. His research will be applied to legal texts compiled by the order of Emperor Justinian (553 CE) with control text samples drawn from Rabbinic (Jewish) law of the same period.

The main objective is to determine whether current industry standards for computational description of legal texts are appropriate to develop a sustainable systematic model with complex reasoning capabilities. The project will consider the opportunities and limitations of theoretical, technological and practical aspects to identify new avenues for legal AI research.

Leverhulme Early Career Fellow Marton Ribary said: “This project builds on the University of Surrey’s strength in historically and theoretically sensitive legal research, utilising the AI expertise present in the School of Law – where new research initiatives address the regulatory, legislative, ethical and philosophical aspects of AI – and the Department of Computer Science – which specialises in AI research with strong links with business and industry. Cross-faculty support guarantees that digital outputs present an innovative and technically sound alternative to current solutions in computational law, with potential industry application.”