Hybrid AI: how knowledge models are essential for trustworthy AI

The session starts with 3-4 introductions of application areas and methodologies that center around Hybrid AI. In an interactive session, the aim is then to identify the main challenges for take-up of hybrid AI. The outcomes of the session will be used as inputs for the H2020 ICT48 projects (specifically TAILOR) that define the roadmap for European AI, and may also be serve as inputs for newly developed Horizon proposals.

The objective of the session is to outline the importance of knowledge models for trustworthy AI. A lot of attention is going to data-driven AI applications, but there are many cases where this problematic due to lack of adequate data. In the light of the proposed AI Regulation by the European Commission and the 7 principles of responsible AI as formulated by the High Level Expert Group, this session will give examples and approaches how hybrid AI contributes to trustworthy applications.