24 October 2024

William Byrne Receives NordForsk Grant of 702.656 NOK

The project

will seek to create benchmarks for testing AI-based legal decision support systems in asylum law in Denmark and Norway

Picture of WIll

NordForsk awards funding to Nordic Refugee Determination Project led by University of Copenhagen

NordForsk has awarded Assistant Professor William Byrne (UCPH), along with Postdoctoral Researcher Runar Lie (UiO) and Prof. Thomas Troels Hildebrandt (UCPH), a grant of 702.656 NOK to support work within the project "Nordic Refugee Determination: Advancing Data Science in Migration Law" (NoRDASiL). The funding is part of the Nordic Programme for Interdisciplinary Research.

The NoRDASiL project is a collaborative initiative, led by Prof. Thomas Gammeltoft-Hansen, focused on improving transparency and fairness in asylum decision-making using data science and computational techniques. The project has developed a substantial database of over 750.000 asylum case files across the Nordic countries, enabling a comprehensive analysis of patterns and inconsistencies in legal outcomes. By integrating insights from law, data science, and social science, NoRDASiL aims to provide policymakers and decision-makers with new tools and methods to better understand and reform asylum processes.

The additional grant supports Work Package A (WP-A) of the NoRDASiL project, which seeks to create benchmarks for testing AI-based legal decision support systems in asylum law in Denmark and Norway. This work addresses the increasing availability of large language models (LLMs) trained in Nordic languages and the rising interest in their application in public administration and law. WP-A aims to develop evaluation tools to test how well LLMs and symbolic models can recognize legal elements, predict reasoning, and infer obligations within asylum decisions. The project will operationalize this by gathering representative legal decisions, establishing a Nordic repository, and benchmarking various models, with findings to be published in interdisciplinary research articles.

Topics