RMS - AI: The Trouble with Large Language Models

We have seen AI produce convincing nonsense - the notorious "hallucnation" This talk looks beneath the hype of ChatGPT and Large Language Models from an engineering perspective. We will explore how these statistical engines actually process information, why hallucinations are a structural feature rather than a simple code bug, how modern tachniques are attempting to constrain them, and how the shift toward "Agenic AI" aims to build reliable systems out of inherently unreliable components

Our speaker, Russell Whitworth, is a memeber of the U3A AI Community.