Large Language Models (generative models of text) power a growing class of products and applications, such as Google Bard or ChatGPT, which seem poised to transform the way we interface with knowledge and build our personal and economic lives around it. In this broadly non-technical talk, I will seek to demystify the technological shift underpinning this paradigm. In doing so, I will explain why this shift is interesting to researchers and downstream users alike, and offer some suggestions as to where this form of technology might be heading next.