With Rafael Pérez y Pérez, I wrote “Computational Models for Understanding Narrative,“ just published! We discuss symbolic models of narrative and how LLMs, while they are intriguing, are not the same and cannot do the same things.
@nickmofo excellent. Very much looking forward to reading this
@nickmofo btw did you see me and @maxkreminski’s in2writing paper on similar things? More from the perspective of computational support for story authors. https://aclanthology.org/2022.in2writing-1.11/
@chrisamaphone @maxkreminski Thanks for this! Good to see this work in computer writing assistance, which I’ll dig into
@nickmofo @chrisamaphone @maxkreminski it's a bit off from this conversation, but I've been looking for collaborator for a literature review on how people use LLMs when writing narrative. The idea is that when one looks systematically at ways to work together with an AI (similarly to what the work by Kreminski and Martens suggest), there are many ways such systems could contribute. As an example consider AI Dungeon. I'd think it's extremely successful, even though it is very weak on narrative structure. Similarly, we've recently looked into (non necessarily AI) collaborative writing systems https://doi.org/10.1007/978-3-031-42286-7_5 using a generic collaborative writing taxonomy and I think there's lots to be learned from redoing it for LLMs
