AI for Peace, Justice and Security

Making AI fair, just and beneficial for all of society, while also keeping the Dutch industry a competitive world player in this segment of technology. This enormous challenge brings many opportunities to the domains related to Peace, Justice and Security (PJS). AI is changing the public domain, because the actions, communications and the associated public/private services become increasingly automated, programmable and intelligent. The use of AI in Peace, Justice and Security should strike the proper balance between new and optimized functionalities while also protecting impacted citizens as well as the institutional foundations of our society for AI vulnerabilities and dependencies.

When looking at (the use of) AI in PJS domains, several aspects are quite distinctive. Firstly, the data within these domains is characterised by its high level of sensitivity, as it is about 'reliability' of persons, of organisations and states, and their behaviour. Another aspect is the character of the data. It is highly asymmetric because the class of interesting cases are always outliers in an ocean of non-problematic data. Another typical characteristic in PJS domains is explainability. In PJS, AI is used in a public governance context, for example in the administration of justice. This puts heavy requirements on explainability AI in PJS domains. Finally, PJS also concerns the AI methods themselves. Such as the justice component (legal responsibility) and the security aspects of the performance of AI (attempts to mislead AI/adversarial learning, or to exfiltrate information from AI systems). All of these aspects generate various interesting, innovative and valuable lines of research, in which different disciplines converge to answer what this means and/or looks like in concrete, both technical and functional.

We see most opportunities in convergence in domains such as cyber security, criminal justice chain, and legal tech. This can be both applied questions: technique-related questions (IN AI), applications towards (cyber)security and the administration of justice (WITH AI), and regulatory questions, e.g., on accountability dynamics, and transparency. Furthermore, fundamental work on AI focuses on forensic data (e.g., Natural Language Processing (NLP) for PJS) and applications, which is specific on many accounts, including the institutional and legal context and the language.