The creation of Leuven.AI, the KU Leuven Institute for Artificial Intelligence, was partly driven by promoting interdisciplinary investigations into artificial intelligence. Besides methodological and applicational research, Leuven.AI dedicates a distinct research line to the philosophical, ethical, legal and societal implications of AI in acknowledgement of the deeply transformative role this technology plays in our societies. In this presentation, we will showcase three recent projects on liability, copyright and open innovation.
The first part reflects on the allocation of liability between actors involved in the supply chain when AI causes damage. This emphasis on liability is not surprising considering that AI systems will increasingly cause damage. Nevertheless, the application of liability regimes for damage caused by AI systems can be challenging. The characteristics of AI systems make it difficult to trace back potentially problematic decisions made with the involvement of such systems. This in turn may make it challenging for victims to obtain compensation under different liability regimes. This part will focus on some challenges that arise when applying liability regimes to damage caused by AI and provide normative solutions to overcome some of the identified issues.
The second part will discuss the role that copyright and connected rights play in the process of training models for machine learning purposes. The discussion will focus on the current status of EU copyright law and will try to identify how the rights under analysis can foster an open, transparent, competitive and trusted AI environment.
The final part will focus on AI from a patent law perspective. A first series of questions will revolve around the eligibility of AI for patent protection and the position of the ‘inventor’. A second series of questions will focus on open innovation models in AI and beyond, and introduce some basic concepts which are fundamental in this discussion. (1) What is open? Reference will be made to the concept of ‘open innovation’ versus ‘commons’; (2) To whom is it open? Attention will be drawn to the taxonomy of openness: one to one, one to some, one to all, some to some, etc. and concepts such as license of right, patent pool and clearing house; (3) How can openness be enforced? Introduction of the concept of inclusive patent.