Dr George Baryannis, University of Huddersfield: Abstract

Recent developments such as the EU AI Act and the AI Opportunities Action Plan announced by the UK PM in January, have created a stronger push towards increasing explainability of intelligent systems. This talk begins with an overview of the current AI landscape, summarising the two main families of approaches, knowledge-based AI (also known as Good Old-Fashioned AI) and data-driven AI (including primarily machine learning). These different approaches will be evaluated through the lens of explainability, contrasting the inherent explainability of knowledge-based approaches with the range of ante-hoc and post-hoc interpretability options for machine learning approaches. Following on from this, some indicative approaches that try to bridge these two worlds will be presented, to benefit from their collective advantages while mitigating their drawbacks. Finally, we will explore how these approaches can be applied in domains related to supply chains, resilience and sustainability.

Organised by BCS SGAI The Specialist Group on Artificial Intelligence http://www.bcs-sgai.org