with apologies for the paywall -- if you'd like to read it but lack access, contact me and I'll forward by email.

We map 17C English along an axis of good-evil to trace contours of early modern gender ideology (fascinating and bizarre in our data) as well as a preview of methods for identifying the most salient moral contrasts that structure the semantic space of any corpus.

LLMs will never be able to make you feel embarrassed that they don't like you. This, in the end, is the real meaning of personhood and the problem of other minds.

hesitate to make predictions, but... seems like once the base models cross a certain threshold, there will be so much to do in terms of building applications that the $$ incentives to train new models will recede, and it'll be back on universities to come up with the next big-$ innovation

We had a visiting speaker come in last month and explicitly say that universities won't be able to keep up bc research in the field is too resource intensive. sounded true, or at least likely. but there's also a huge middle ground between building new LLMs and working with students

that's tricky. I have always struggled getting majors to enroll, and for my nonmajors the literature presents more obstacles than I expected. after several false starts I rebranded the course "Math 4 Shakespeare" (just what it sounds like) and have gotten great responses from nonmajors. good luck!

undergrad or grad? are you more training future colleagues who might write quant theses or more offering a fun interdisciplinary class to a bunch of future doctors and engineers?

I’ve been there and had the exact same realization

Yes but only bc authors are cagey af and too clever by half. Honestly I’d trust an LLM to represent the ideas and intentions accurately over abt 90% of authors 90% of the time. Like if you ask me about my book I’ll spout mostly nonsense