“Explainable artificial intelligence means the reasoning behind a decision should be comprehensive for its human users.”
Research focus: Artificial Intelligence, application of AI techniques to social issues, data science
Floris Bex is computer scientist at Utrecht University. He studies how people reason, how this reasoning can be captured in formal models and how it can be supported and improved using Artificial Intelligence (AI) technologies. He is involved in argumentation theory in AI and AI & Law. He is also active as Professor at the Tilburg Institute for Law, Technology, and Society (TILT) of Tilburg University.
National Police-lab AI
Floris is the scientific director of the National Police- lab AI, a cooperation initiative of the Dutch National Police with several universities. The lab aims to develop state-of-the-art AI techniques to improve the safety of the Netherlands in a socially, legally and ethically responsible way. The lab is a member of the national ICAI network.
The research of the Police Lab location Utrecht focuses on reasoning with (legal) arguments, explainable AI and the integration of symbolic and subsymbolic techniques within AI. Together with the ALGOPOL project, it also researches how AI can be used responsibly in the police force.
Explainable AI system to assess crime reports
Floris and the lab have designed an explainable AI system to assess crime reports more efficiently. "Only a small percentage of civilian complaints are accepted by law as criminal offences. If an automated system does not explain why a complaint has been rejected, it can undermine trust. That's why we designed a crime reporting tool with a chatbot that reasons with legal arguments so citizens get an understandable and legally sensible explanation of why their report is not being processed."
Areas of interest:
- Socially, legally and ethically responsible development of AI techniques
- Text & data analytics
- Computational argumentation
- Multi-agent systems
- State-of-the-art AI to improve safety in the Netherlands