Much of today’s artificial intelligence (AI) revolution is powered by data-driven techniques like large language models. While this approach has proved to be tremendously successful, the way in which these systems reach decisions are not always transparent to human observers. The Chair in Symbolic Artificial Intelligence works at the forefront of an alternative tradition: logic-based AI, which uses formal reasoning to address these and other problems related to the data-centric approach. Based in the Department of Computer Science and part of UCT’s AI Research Unit, the Chair is one of the main drivers worldwide of a research area referred to as Knowledge Representation and Reasoning (KR).

The Chair’s particular expertise is in the sub-area of KR known as Non-Monotonic Reasoning, which is concerned with how an intelligent agent – human or artificial – should revise its conclusions when new – and possibly conflicting – information comes to light. The Chair has championed a comprehensive framework for tackling this challenge, encompassing the formal logical languages used to represent knowledge, the techniques for reasoning over them, the properties those techniques should satisfy, and the algorithms needed to implement them efficiently. Recent contributions include adapting existing reasoning techniques to more expressive logical formalisms, broadening their applicability, developing methods for improving usability for human knowledge engineers and laying the groundwork for more efficient algorithms. Together with other international leaders in the field the Chair has also led a survey paper intended to serve as a roadmap for KR, contextualising the strengths and challenges of this approach at a moment of intense global attention on AI.