Ian Miell
- talk
A Week's Work in Six Hours: How LLMs will (Probably) Revolutionise the Study of Humanities… But Not Yet
Learn from Ian" I set myself a challenge, which was to try and write a high quality essay using LLMs to help me research the subject in under two days. And I had two rules: One was I couldn't look at my original essay from 1996; and I couldn't read the texts that I was set to read for this essay"
This talk shares my experience using both private and public Large Language Models (LLMs) to write an undergraduate history essay that I originally tackled nearly 30 years ago. The essay, which explores the question of "How far was New Liberalism a departure from the Old Liberalism?", serves as a case study to examine the impact of AI assistance on historical research and essay writing.
There were rules to this task: No peeking at the original essay No reading any of the suggested material, except to find references.
To achieve this I used various tools, including ChatGPT, Llama2, an old Dell Xeon tower PC, some python and good old-fashioned grep. The talk will cover what I learned about how humanities study is both being revolutionised by LLMs, and how in many ways nothing has really changed. It will also cover what I learned after I publishing a blog post on it that made the front page of HackerNews from others about how complacent faculties around the world are about what's coming.
