The following is material from a talk I gave. For a panel at MLA 2024 on LLMs. Editing it, because I addressed too many different topics in one talk, and I want to do each topic a little more justice. Many thanks to Melanie Walsh for organizing & facilitating it!
Three years ago, I finished my PhD in English Literature. Since then, I’ve been mutating into…something else — via a few different “Machine Learning Research Scientist” positions. That title still feels very foreign to me, but it has been a fruitful time.
Given that this kind of transformation is still, relatively, unique within English departments, I wanted to take the opportunity to give my perspective on Large Language (or Multimodal) Models, English Departments, and the Content Generation Industry.
In the future, I’ll take some time to write about things I’m reading in the field. (And, if I’m feeling up to it, I’ll try to write how someone in literary history might run with the idea.) For now, I just want to take the time to answer some questions that I frequently get from colleagues in English departments. It’s increasingly apparent that there’s momentum to critically think about these technologies.
What I’ll provide for now are my initial responses, some anecdotal experiences to explain them, and caution that my impressions are to be taken with a hefty dash of salt.
Part 1: Should Students in English learn to train & fine tune language models?
(In Progress) Will Professional Content Creators’ Reception of ML Change in a Few Years?