I'm..well...not skeptical of the power of AI, but I'm skeptical of the newness of some of the things its proponents claim. The field of digital humanities has been around for ten or twenty years, at least, and they've been doing similar work to that described in this article for all that time. Maybe the programs they were using could be described as AI, or maybe the development between those programs and the ones we now consider AI is simply an incremental one that has received a huge amount of hype generated by the publicists of whoever owns ChatGPT. It's hard for me to judge.

There have been researchers using computers to look at authorship for many years. (Shakespeare is the big one, but there are others.) I feel pretty sure that the programs that have been doing such work for years could also do the particular project described here, if they thought the metric of story-time-per-page was important. I've read a cr*p-ton of literary criticism over the past five years while I was working on a PhD I don't think I've told y'all about (I should defend in late spring), and none of those writers have had anything to say about the passage of time per page.

Are there scholars who are deeply interested in this? Surely. But I'm not convinced that tools to look for it didn't already exist, nor am I convinced that somebody wasn't already doing this work, as it seems a straightforward application. The recent breathless hype over AI seems to claim that it's something all-new, when it seems more like an interface that makes it more convenient to ask computers questions that they could have been asked before, albeit in a way that might require a little more expertise of the asker.

In my own field, digital humanities scholars showed pretty conclusively years ago that Agatha Christie's later books show changes in the language used that suggest that she was developing dementia. That's interesting work that was better, quicker, and more easily done with a computer than sitting with her books for years and doing it by hand, but the program that did it was not called "AI." (Ay least I think it wasn't.) It was probably just some kind of fancy database of its time, which was in turn some kind of fancy concordance of its time.

As a writer and teacher, I'm very interested in the ways that writers handle time in their narratives. I learned years ago that it was possible to stop time with words when I wrote a scene in which bullets are flying but it takes (if I recall correctly) a couple of pages for them to land, because I spend a few paragraphs in each player's head as they decide how to respond to the crisis, and each of their decisions is based on the best interests of the person they love most. For one of those people, that person is herself.

If I were to work on the passage of time in narrative, I would choose some scenes like this one (only not one written by me) that handled time interestingly and consider it closely. If I used a tool like the Ai described here, I would use it as a comparison for the scenes I'd considered with an understanding of what the author was trying to do. Is this scene typical of their handling of time? Is it typical of fiction written by the author's contemporaries or is it something more rare or even unique?

Questions like those are interesting to me. Reporting the average number of hours per page over a large variety of books without considering what that means tells me nothing. And I really doubt that this is such a widely vexing problem to literary scholars that it justifies this article's thesis that it "profoundly bothers the kind of people who study literature."

quote:

Reporting the average number of hours per page over a large variety of books without considering what that means tells me nothing.

This is the crux of the matter, IMO, the heart of what is missing with AI. And, perhaps most importantly, the reason why I agree with those who say that the "i" of AI is a misnomer, because there's no "intelligence" on the part of the AI-tool being used.

The current AI tools available are not capable of considering what something means. Whether they ever will be is a different question.

But for now, AI is a tool and what really matters is how humans use it...

And from what I can tell, that's the stance of UIUC's Underwood as well.