Mike Livermore speaks with ICA4 Fellow Deshen Moodley, Associate Professor in the Department of Computer Science and Co-Director of the Centre for Artificial Intelligence Research at the University of Cape Town. Moodley’s work focuses on machine learning and probabilistic modeling systems.

Moodley begins by remembering his first encounter with computers, and how that spurred a lifelong interest in programming and, eventually, led him, first, to a career in the private sector before transitioning to an academic career. Moodley compares the differing aims of industry and academia, in addition to pointing out the way in which the structure of a country’s academic system influences the production of knowledge and the interaction between academia and industry (:55 – 8:53). Moodley then explains what machine learning is, where it fits within the broader field of artificial intelligence, and, after providing a basic introduction to Bayesian inference, offers examples of knowledge representation systems that the public might be familiar with in their daily life. Moodley then explains how these systems can be used to increase efficiency in areas like medical diagnoses and the subsequent relaying of that information to the patient (9:00 – 26:19). The conversation then delves into Bayesian networks, what they are, and how they are used in artificial intelligence research. Moodley describes how AI, using machine learning, might go about diagnosing an illness, before emphasizing the benefits of a hybrid system that combines Bayesian and knowledge-based systems (26:30 – 45:33). Shifting to one of Moodley’s key concerns about AI, the conversation touches on whether machine learning has increased access to knowledge. Moodley argues that fear of autonomous AI is one of the major hurdles to the field, with the result that society has been slow to incorporate these systems (45:45 – 55:22). The conversation concludes with a thought-provoking hypothetical (featuring cameos from Brad Pitt and Angelina Jolie!) illustrating the difference between normative- and performance-based approaches to artificial intelligence (45:45 – 1:08:40).

The Intercontinental Academia (ICA) is a global network of future research leaders sponsored by the University-Based Institutes of Advanced Studies. The ICA4 explores the complementarities between artificial intelligence and neuro/cognitive-science.