Researchers from the Faculty of Law at Vilnius University (VU), together with international partners, have been awarded funding under the NordForsk call “Responsible Use of Artificial Intelligence” and will implement the international project “NordAId: Trustworthy AI in Public Decision Making”. The total project budget amounts to approximately EUR 1.4 million. The project will run for four years and is implemented by a consortium coordinated by the University of Oslo, with additional contributions from researchers at the University of Copenhagen, Uppsala University, and Vilnius University.
The VU Faculty of Law team consists of Assist. Dr Inga Martinkutė, who will lead the university’s research group, Prof. Dr Jurgita Paužaitė-Kulvinskienė, Senior Researcher Dr Monika Žalnieriūtė, PhD candidate Goda Strikaitė-Latušinskaja, and Assist. Dr Neringa Gaubienė. The project will also contribute to strengthening cooperation with the Danish Centre for Artificial Intelligence (CAISA), the Norwegian TRUST-AI Centre, and the developing Lithuanian AI Factory (LitAI), coordinated by VU.
The project aims to develop an interdisciplinary framework for the application of artificial intelligence (AI) in public administration decision-making in the Nordic and Baltic countries. It will assess both the potential and the limits of AI, particularly in complex and sensitive domains. This study focuses on asylum decision-making, where several use cases of AI fall under the high-risk categories defined in the EU Artificial Intelligence Act (e.g., risk assessment, analysis of evidentiary reliability, identity verification systems).
Automated decision-making is highly progressive and useful, especially in crisis situations when the number of standardised applications (for example, asylum applications) suddenly increases. However, the law still requires that such cases be examined or information collected individually, emphasizes VU team leader Assist. Dr Inga Martinkutė. “AI tools will inevitably become a part of efficient administration; however, they also entail significant limitations and risks. With this project, we aim to identify and define these boundaries — that is, in which frameworks and situations, and under which presumptions, automated decisions could be justified, what risks arise, and how such decisions align with legal standards and international conventions,” goals of the project explains the specialist.
One of the project’s foundations is a unique database of nearly one million asylum decisions, created under a previous NordForsk project. This database enables the application of advanced data science methods while ensuring maximum limitation and protection of personal data. Building on this infrastructure, the project will develop, test, and apply AI models designed to evaluate decision-making consistency, potential biases, and other indicators of trustworthiness.
Researchers at the VU Faculty of Law will analyse the legal and ethical foundations of AI use, the implementation aspects of the EU AI Act, and the requirements of transparency and accountability. The project will produce interdisciplinary recommendations for institutions and policymakers, aiming to ensure that AI in the public sector is deployed responsibly and in a manner aligned with the protection of human rights.