What is it like not to be a Large Language Model?

A lecture by Professor Peter McDonald (Professor of English; Fellow of St Hugh's College, Oxford).

As a meta-pedagogical exercise, this talk cuts through the hype and panic surrounding GenAI to confront what Large Language Models are, and are not. It situates them within a longer intellectual history, from statistical linguistics to transformer architectures,
and describes their operation plainly: systems that generate fluent text through probabilistic patterning rather than intention, understanding, or lived experience.

Drawing on rhetoric and the philosophy of language, the talk reopens the language-thought question and highlights a decisive distinction too often obscured in public debate: the key difference lies not in what LLMs can produce, but in how and why they produce it and what that implies for us.