Uses machine learning to study literary imagination, and vice-versa. Likely to share news about AI & computational social science / Sozialwissenschaft / 社会科学
Information Sciences and English, UIUC. Distant Horizons (Chicago, 2019). tedunderwood.com

@tedunderwood.com
Uses machine learning to study literary imagination, and vice-versa. Likely to share news about AI & computational social science / Sozialwissenschaft / 社会科学
Information Sciences and English, UIUC. Distant Horizons (Chicago, 2019). tedunderwood.com
Wrote a short piece arguing that higher ed must help steer AI. TLDR: If we outsource this to tech, we outsource our whole business. But rejectionism is basically stalling. If we want to survive, schools themselves must proactively shape AI for education & research. [1/6, unpaywalled at 5/6] +
I never know when I’m joking anymore
Let's just posit one.
Some of you prob think "I'm skipping the predictable discourse. But—since my sacrifice is silent—no one will ever see or appreciate it." This is why the internet needs to create a Divine Being who can see every act of self-restraint and give it, if not karma points, at least a quiet nod.
Appreciation for beauty? Does … not … compute! Emotions … are … not … logical. System overheating.
Another way of telling the story would be that we had been trying at least since Markov models were applied to Pushkin — 1913.
The short version is, “largely.” The ppl I describe in that piece (e.g. Barthes and Foucault) are not doing philosophy of mind and don’t have the goal of proving, e.g. that wordless thought is impossible. Their goal is more to upset assumptions about the relative importance of thought & expression.
Omg how did I not know that track? I’m going to be humming it now whenever I play Call of Cthulhu
"In a just world, every article about GPT-4 would nod toward Barthes and Foucault." Downing spoonfuls of mayonnaise as I read Ted's essay, because it is too spicy (link in quote below)
In case actual students from 2050 read this: don’t trust Bluesky pundits; consult the primary sources
“Okay, so GLM-1 was a language model but GLP-5.3 was a drug. KLM was an airline; slogan ‘KLM are going to rock you, uh-huh uh-huh’.” — Desperate History majors in 2050 studying for their turn of the century exam.
The best alternative is "I work with text elementals."
that's also the story I had heard
I actually like this. Let’s make fetch happen.
I'm fine calling LLMs cybernetics if y'all are, tbh I like this better than AI anyway
3 + 8 = __ Our revels now are ended. These our actors, As I foretold you, were all spirits, and Are melted into air, into ____ ___. We can fill in those blanks because symbols are semi-predictable, but that's not bad. It's the point of symbols?
lol vitruvian man is adult content
My hypothesis is that over the last 2-3 years, students have had to do a lot more timed in-class writing than they used to do.
Anyone else notice that students’ handwriting is like 2x clearer than it was two years ago? I never stopped giving paper exams, so I have a longitudinal sample — and this is as clear as I’ve ever seen it.
To be fair, it’s extremely hard to motivate new questions and validate new methods at the same time, in like 11 pages; we need more people doing this and more discussion re best practices
That was also my response to a couple