Has AI come up with any interesting analyses in the humanities?

[NewYearNewNameWhoKnew]

Inspired by comments about AI use in humanities PhDs - has anyone tried to analyse humanities topics with AI and has it said anything new and interesting? I've heard some much more rapid hypothesis generation in sciences- has it come up with any new thoughts on Shakespeare or similar? Any novel answers to philosophical questions?

I've not been impressed by any Shakespeare responses (all boring, at best GCSE level). I've come to the conclusion it still takes a human to do that, and even with actual humans it's very rare you come across anthing original and convincing anymore anyway.

I have a Philosophy colleage who is totally enthralled by AI, however. It all went over my head a bit, but he's having role-play dialogues with his AI, and it's blowing his mind; I'm sure he mentioned something along the lines of asking it to speak in its 'shadow side', Jungian syle, or something similar. And asking the AI to examine itself in various ways...sounded interesting.

In my field it just comes up with trite and often factually incorrect stuff.

Ooh, I want in on this thread too! (The humanities PhD other one is mine.) Great question. So far I’ve only used AI to input essay titles/topics to compare with what students are writing, apart from once asking Chat GPT to write a short story as if by an author of a certain demographic. What it wrote was weirdly “good” - cliched, but actually kind of believable, at least on first reading. To be honest it freaked me out thinking about what happens to literature if a computer can seemingly create it, so I haven’t ventured back since!

I got it to write something in the style of my favourite author and what it produced read as a horrible pastiche.

Well, that’s reassuring I guess?!

What I was really imagining was feeding in historical sources and seeing if AI comes up with something novel - what was Stonehenge for, or was Shakepeare really a single person, was Elizabeth the first a virgin or not - those kind of historical arguments or just throwing in a load of archeological data and seeing what AI comes up with (you can probably tell I'm a scientist so don't have good questions to ask)

I asked AI a very niche question in my humanities field and it came back with an answer that was pretty much word for word something I wrote…as I’m the only scholar to have linked the two things. Not impressed! I have colleagues who rave about its capacity to, eg, count active nouns in Romantic female authors’ texts, which they claim reshapes our understanding of female authorship and passivity etc…not really convinced.

Nope

Shame, you'd think just from a data crunching point of view it would have the capacity to surprise us- link some obscure artefacts in Belgium with an art work in Rotherham and thus discover an ancient trade route no one ever suspected - or solve all the episodes of Fake or Fortune so we don't have to watch them.

But that would only happen if it had access to those sources and (most likely) you pointed out their relevance.

I used ChatGPT a lot (but not for humanities research!) and it's ability to ignore even the most obvious things unless you explicitly tell it not to is astounding. It goes down rabbit holes. It mis-interprets the guidance you give it, or rather it fails to see any nuance. I'd be amazed if it could come up with the nuance dot-joining that is required to make new 'discoveries' or interpretations in the humanities.