Maria Antoniak

It definitely would! On my mind also because I was just reading how Emily Bronte edited different versions of Wuthering Heights that "dilute" some of the accents for her non-Yorkshire readers.

Excited to have the Big Picture workshop back for another iteration this year at @aclmeeting.bsky.social
Submit your big picture ideas, consolidation work, phd thesis distillation, etc. by March 5th!

www.bigpictureworkshop.com
w/ Allyson Ettinger, @norakassner.bsky.social, @sebruder.bsky.social

I'll be in NYC for a visit soon and this is going to be my first stop. Very persuasive photo.

The unintended consequences of monitoring technologies: Evidence from the Electronic Logging Device mandate The electronic logging device (ELD) final rule was passed in 2015 and implemented in phases over the course of four years. The mandate required that m…

Interesting new statistical analysis, again confirming that mandatory electronic monitoring in trucking—ostensibly a safety measure—led to *increased* accident and fatality rates after implementation. www.sciencedirect.com/science/arti...

Doo do doo

This is probably the part of the faculty job that scared me the most, and it’s still a bit scary, but it’s a much more collaborative and imaginative process than I pictured!

“ARC Raiders remains as one of the most popular games on Steam, as it has been in the top 5 of the global weekly sellers on Steam for the last 12 weeks” wow congrats!

So interesting!!

We find:

1️⃣ Stories in mid- and low- resource langs have more misreps

2️⃣ Stories from tier-2 and 3 regions had more factual and cultural misreps

3️⃣ Most misreps are about social norms, practices, & food

Lots of other analysis in paper! Clear avenues of improvement!

6/10

🎭 How do LLMs (mis)represent culture?
🧮 How often?
🧠 Misrepresentations = missing knowledge? spoiler: NO!

At #CHI2026 we are bringing ✨TALES✨ a participatory evaluation of cultural (mis)reps & knowledge in multilingual LLM-stories for India

📜 arxiv.org/abs/2511.21322

1/10

LatinCy Readers logo

Announcing—LatinCy Readers v.1.0.2, i.e. LatinCy-powered corpus readers for Latin text collections. Quickly get sentences, lines, words annotated for lemma, POS, morphology, NER, etc. Supporting .txt, .xml, .tess, .conllu, and more. github.com/diyclassics/... #digiclass #nlproc

I'm a huge Hollow Knight and Silksong fan :)

Inside the marketplace powering bespoke AI deepfakes of real women New research details how Civitai lets users buy and sell tools to fine-tune deepfakes the company says are banned.

A study from Stanford and Indiana University reveals how the Civitai marketplace facilitates the creation of nonconsensual AI deepfakes.

Backed by Andreessen Horowitz, the platform allows users to post "bounties" for custom AI instruction files, known as LoRAs, designed to mimic real individuals

Yes I've been very careful! No spoilers!

That makes me so excited! I'm in love with Aloy and her world!

A screenshot of the remastered Horizon Zero Dawn game, showing a woman with long red hair riding a robot animal in an alpine valley.

I bought myself a PS5 and have been obsessively playing Horizon Zero Dawn. It's SO good! Red Dead Redemption is next on the list, but I've never owned a PS before and would love any recommendations. I like beautiful stressful games.

CMU LTI Summer 2026 Internship Program Application We are looking for applicants for the Carnegie Mellon University Language Technology Institute's Summer 2026 "Language Technology for All" internship program. The main goal of this internship is to pr...

🚀 Apply to CMU LTI’s Summer 2026 “Language Technology for All” internship! 🎓 Open to pre‑doctoral students new to language tech (non‑CS backgrounds welcome). 🔬 12–14 weeks in‑person in Pittsburgh — travel + stipend paid. 💸 Deadline: Feb 20, 11:59pm ET. Apply → forms.gle/cUu8g6wb27Hs...

Not sure whether I agree, but a language model invoking Piranesi to argue w/ me about memory and personhood is an automatic RT

It took me weeks, but finally it's there: an overlong blogpost on synthetic pretraining. vintagedata.org/blog/posts/s...

The metaphor of the broken staircase emphasizes the absurdity of creating elaborate workarounds for the "missing stair" rather than directly "fixing the staircase."

This was certainly part of the challenge with Epstein; pretending everything is fine and everyone should remain part of the community.

Relevant:

Beware the "missing stair" — "a person within a social group who many people know is untrustworthy or otherwise has to be 'managed', but whom the group chooses to work around, by trying to quietly warn others of their behavior, rather than deal with them and their behavior openly."

If you're reading about Epstein and his friends today, could you also take time to read this document?

I guarantee that someone at your university right now is a sexual harasser and/or abuser. Maybe you could stop them.

I also had a too ambitious plan to survey people on ambiguous examples and correlate their labels with their expertise… but I don’t have time for IRB right now.

Yeah I think this is very community specific (like NeurIPS requiring that exact wording in position papers while the idea of a “position paper” being unintelligible to other communities) but could be fun to look at!

All that is to say, we could try relabeling to separate position and review papers, and I might even try it, but it’s not a simple set of decisions for this diverse dataset.

Of course some cases are very clear, especially when considered within a specific discipline’s norms. And we tried to label according to traditional “quant CS” norms, to mimic what we expected of the moderators.

We didn’t quantify that, and while we could try to separate them, part of what we argue in the paper is that all these “paper types” are hard to disentangle. Especially for e.g. FAccT-y work, a single paper could be part review, position, qual, and quant, and readers may disagree on those labels.

Again I'm not saying anything about intent, so far it looks like a difference in conversational style. But I don't think I'm seeing what you're seeing in the video, at least in those 5 min.

Parts of both the slur and sentience discussion felt that way to me.

Sometimes it's not exactly putting words in mouths but rather dropping a conversational bomb (how to determine sentience) that would potentially form the grounding of the conversation if Emily didn't interrupt.

I don’t know if there’s bad intent, but I’ve definitely met people with this kind of communication style, and it’s not fun. Caveat that yeah, I haven’t watched the whole thing.