Could we build a better justice system using artificial intelligence? Could we trust AI to decide legal cases? Too often, justice is not served. Delays leave victims, witnesses and the accused waiting years for a decision on their case.

Even if algorithmic decision-making could help with this delay, public trust in human-computer interaction in justice is under-explored. Can public confidence be maintained in a move to automation? To help answer these questions, our work uses an interdisciplinary approach, informed by legal analysis and empirical data, to produce a model that can decide and sentence a traffic offence.

We incorporate methods from the social sciences to build mathematical models in a collaborative way. Over the next year, members of the public, judges and lawyers will interact with the system and give detailed feedback so that we can understand how automated decision-making and AI might genuinely help to improve justice.

In collaboration with Vivienne Griffin from Somerset House Studios, a striking 3D game environment, MERCY, has been created to convey the research in an accessible way.

Project Lead

Cari Hyde-Vaamonde
The Dickson Poon School of Law, King’s College London

Funder

Economic and Social Research Council (awarded through the London Interdisciplinary Social Science Doctoral Training Partnership)
MERCY project funded by King’s x Somerset House Studios collaboration scheme

Acknowledgements

Pierpaolo Vivo
Department of Mathematics, Faculty of Natural, Mathematical & Engineering Sciences, for support in mathematical modelling

Benjamin Bowling
The Dickson Poon School of Law, for support and feedback

Vivienne Griffin
conceptual artist, with whom MERCY visualisations were produced

Anonymous respondents
involved in research & model design