Webinar on ChatGPT

Heroes and horror stories: hands-on advice for dealing with ChatGPT and its soon-to-be-unleashed cousins presented by Martin Bekker.

There are several myths about ChatGPT, the large language model-based machine learning tool that has upended our world. The first myth is that it can be outwitted, detected or stopped; however, it will keep improving and is likely to be exponentially better by the end of 2023. The second myth is that it can "do anything," but it is simply a pattern-matching exercise and does not comprehend or understand, with several limitations-by-design. The third myth is that it spells the end of the essay or coding problem set, but it shouldn't, as it remains important to teach skills that computers cannot do. This talk will explore and explode these myths, but also discuss ways in which the use of ChatGPT has already revolutionised learning and teaching, but also reflect on where (and how) its use can go horribly wrong (practically, but also ethically). 

Martin Bekker is a computational social scientist and AI ethics researcher, and lectures ethics and social science at the School of Electronic and Information Engineering. Bekker holds degrees in values studies and philosophy from Stellenbosch, Peace Studies from Bradford, Development from the London School of Economics, and a PhD from UJ. His PhD thesis was a quantitative analysis of popular protest in South Africa, using machine learning and protest event analysis. Before joining the academy, he was head of strategy and later of research at the Royal Bafokeng Administration, and served on the World Economic Forum’s advisory panel for sustainable mining and minerals. Bekker was also part of the National Lockdown/Democracy survey team, administered by UJ and the HSRC.

Add event to calendar