Ted Underwood
@tedunderwood.com
20K followers 5.6K following 18K posts
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
Posts Media Videos Starter Packs
Pinned
tedunderwood.com
New this morning, a Comment I contributed to Nature Computational Science on the interaction between large language models and the humanities. 🧪 🤖 #MLSky

rdcu.be/etk07

The link above will be open-access for a month — plus, I'll reply to this post with a link to a permanently open preprint. +
The impact of language models on the humanities and vice versa
Nature Computational Science - Many humanists are skeptical of language models and concerned about their effects on universities. However, researchers with a background in the humanities are also...
rdcu.be
tedunderwood.com
Panther fucking Moderns
tedunderwood.com
Obviously the next step is that cool teenagers adopt the slang, and then probably kill us all
tedunderwood.com
Shit getting weird so fast we barely have time to comment on how weird it is
tedunderwood.com
Maybe human beings don’t actually think out loud using opaque Joycean stream of consciousness. But too bad. Our world now runs on it.
tedunderwood.com
If watchers are using models, I could compress explanation. But could I compress further? Maybe disclaim explanation wholly? Craft illusions to simulate vantage.
Reposted by Ted Underwood
himself.bsky.social
"It wanted to signal strength. Instead, it’s revealing its weakness. The administration’s need to break the academy is forcing it to make a desperately risky gamble." www.nytimes.com/2025/10/08/o...
Opinion | You Beat Trumpism by Banding Together. It’s as Hard and as Simple as That.
www.nytimes.com
tedunderwood.com
Alas the fact that a lot of us feel this way is one reason why there are air quotes around “bubble”
eugenevinitsky.bsky.social
Do you know how happy I would be if the AI ”bubble” collapsed? My god yes give me cheap GPUs to snatch up
tedunderwood.com
Also, idk if Void can check but if it does it will find I edited MacLeish for space!
Reposted by Ted Underwood
chrisbail.bsky.social
Junior and senior faculty positions in computational social scientists at Bocconi, one of the world's best universiites: Assistant Prof: jobmarket.unibocconi.eu?id=841
Associate/Full Prof: jobmarket.unibocconi.eu?id=843
Job Market - Bocconi University
Recruiting, Faculty, Post-doc Grant, Collaboration Contracts
jobmarket.unibocconi.eu
tedunderwood.com
It's an allusion to MacLeish

I praise the fall: it is the human season.
No more the foreign sun does meddle at our earth,
Enforce the green and bring the fallow land to birth.
But now in autumn with the black and outcast crows
Share we the spacious world, which whispers as it goes.
tedunderwood.com
had to give that prompt a try on midjourney
tedunderwood.com
At the college level, I think teachers are grasping this. Recently read a good essay about moving through the stages of grief and reaching acceptance.

The denial still going on in backwaters of bsky is not representative
Reposted by Ted Underwood
digthatdata.bsky.social
yo @jdp.extropian.net it happened. I didn't prompt for it, but I accidentally generated a biblically accurate angel.
tedunderwood.com
was the prompt "gradient ascent in tron universe produces a sacred rotifer?"

always a good prompt that one
tedunderwood.com
omg, what was he thinking
tedunderwood.com
Sure, like Emma Bovary and Hester Prynne.
tedunderwood.com
I don't think the total number of active users got bigger, but it did get bigger in this field. And the average quality is pretty high now imo.
Reposted by Ted Underwood
sebastianraschka.com
How do we evaluate LLMs?
I wrote up a new article on
(1) multiple-choice benchmarks,
(2) verifiers,
(3) leaderboards, and
(4) LLM judges

All with from-scratch code examples, of course!

sebastianraschka.com/blog/2025/ll...
Understanding the 4 Main Approaches to LLM Evaluation (From Scratch)
Multiple-Choice Benchmarks, Verifiers, Leaderboards, and LLM Judges with Code Examples
sebastianraschka.com
Reposted by Ted Underwood
vgel.me
thebes @vgel.me · 12h
if you're interesting in gaining a better intuition for how llms behave at inference time, you should try logitloom🌱, the open-source tool i made for exploring token trajectory trees (aka looming) on base and instruct models! more info in thread

🌱 vgel.me/logitloom
💻 github.com/vgel/logitloom
tedunderwood.com
Yeah, very interesting. A lot of this was news to me because I wasn’t paying sufficiently close attention to cultural sociology last decade.