Lightnews — Scholar-powered news

Maria Antoniak

@mariaa.bsky.social

10K followers 1.3K following 2K posts

☀️ Assistant Professor of Computer Science at CU Boulder 👩‍💻 NLP, cultural analytics, narratives, online communities 🌐 https://maria-antoniak.github.io 💬 books, bikes, games, art

maria-antoniak.github.io

Posts Media Videos Starter Packs

Pinned

Maria Antoniak @mariaa.bsky.social · Nov 20

some little bluesky tips 🦋

your blocks, likes, lists, and just about everything except chats are PUBLIC

you can pin custom feeds; i like quiet posters, best of follows, mutuals, mentions

if your chronological feed is overwhelming, you can make and pin make a personal list of "unmissable" people

15 53 240

Reposted by Maria Antoniak

Mel Andrews @bayesianboy.bsky.social · 3d

What problem is explainability/interpretability research trying to solve in ML, and do you have a favorite paper articulating what that problem is?

15 9 50

Maria Antoniak @mariaa.bsky.social · 5h

"come on pleeeease we only want to persuade you of good things! please please just let us persuade you a little bit! we promise only for GOOD things! come on!"

Maria Antoniak @mariaa.bsky.social · 5h

"nlp for social good"

Marjorie James Keenan @unenthusiast.com · 11h

In honour of spooky month, share a 4 word horror story that only someone in your profession would understand.

rm -rf ~/

hammancheez @hammancheez.bsky.social · 11h

"The chancellor approved it"

1 1 28

Reposted by Maria Antoniak

David Bau @davidbau.bsky.social · 1d

What does an LLM do when it translates from Italian "amore" to Spanish "amor" or French "amour"?

That's easy! (you might think) Because surely it knows: amore, amor, amour are all based on the same Latin word. It can just drop the "e", or add a "u".

2 2 33

Maria Antoniak @mariaa.bsky.social · 1d

Lots of funny results along these lines also in this paper by Alan Saji: aclanthology.org/2025.finding...

RomanLens: The Role Of Latent Romanization In Multilinguality In LLMs

Alan Saji, Jaavid Aktar Husain, Thanmay Jayakumar, Raj Dabre, Anoop Kunchukuttan, Ratish Puduppully. Findings of the Association for Computational Linguistics: ACL 2025. 2025.

aclanthology.org

1 1 10

Reposted by Maria Antoniak

Caitlin Rivers @cmyeaton.bsky.social · 1d

For the 2nd week in a row, the federal shutdown has blocked disease surveillance reporting. So Team Force of Infection again visited all 50 state websites to track COVID, flu & RSV. caitlinrivers.substack.com/p/outbreak-o...

Outbreak Outlook: Week 2 of DIY Surveillance

State by state disease surveillance of COVID-19, influenza and RSV amid federal public health cuts

caitlinrivers.substack.com

2 27 72

Reposted by Maria Antoniak

Mor Naaman @informor.bsky.social · 1d

This post violates my pet peeve -- don't list a date without a day-of-week if you want people to sign up -- but this protest is the most important thing you can do this coming SATURDAY.

You've posted on Bluesky enough. We need everyone on the streets. #nokings

Timothy Snyder @timothysnyder.bsky.social · 3d

No Kings. October 18th. Free speech is action. Please join in.
www.nokings.org

No Kings

As the president escalates his authoritarian power grab, the NO KINGS non-violent movement continues to rise stronger. We are united once again to remind the world: America has No Kings and the power ...

www.nokings.org

5 8

Maria Antoniak @mariaa.bsky.social · 2d

The #COLM2025 workshop on NLP4Democracy is starting now! Join us in 520E.

I’ll be speaking at 10:15am with @ysiglidis.bsky.social about work with @iaugenstein.bsky.social and @serge.belongie.com focused on tracking collective narratives on social media.

A slide that reads “NLP and all the interesting and weird ways it intersects with processes and values that comprise democracy”

5 19

Reposted by Maria Antoniak

Melissa Terras @melissaterras.bsky.social · 2d

Job! Postdoc in Digital Cultural Heritage: Data Ethics & Software Ecosystems, on new EU funded grant. 3.5 years, @designinf.bsky.social, Edinburgh. Design ethical processes, & evaluate open source tools for Digital Cultural Heritage Objects, with me! elxw.fa.em3.oraclecloud.com/hcmUI/Candid...

Postdoctoral Research Associate (Digital Cultural Heritage Data Ethics & Software Ecosystems)

We are looking for a Postdoctoral Research Associate in Digital Cultural Heritage Data Ethics & Software Ecosystems to join the Institute for Design Informatics, Edinburgh College of Art.

elxw.fa.em3.oraclecloud.com

37 33

Reposted by Maria Antoniak

Sunnie S. Y. Kim ☀️ @sunniesuhyoung.bsky.social · 2d

Our Responsible AI team at Apple is looking for spring/summer 2026 PhD research interns! Please apply at jobs.apple.com/en-us/detail... and email [email protected]. Do not send extra info (e.g., CV), just drop us a line so we can find your application in the central pool!

Machine Learning / AI Internships - Jobs - Careers at Apple

Apply for a Machine Learning / AI Internships job at Apple. Read about the role and find out if it’s right for you.

jobs.apple.com

2 11 24

Maria Antoniak @mariaa.bsky.social · 2d

Closing session for #COLM2025!

There will be #COLM2026! @yoavartzi.com and @gregdnlp.bsky.social will be organizing. Location TBD.

Full day of workshops tomorrow, check the program.

1 20

Reposted by Maria Antoniak

David Mimno @dmimno.bsky.social · 3d

He had physical copies of at least six books at the podium and read passages from them. His main point was that regardless of whether you're more concerned about Big Risks or current harms, the "everything is great" position is untenable for anyone.

1 1 8

Reposted by Maria Antoniak

Thomas Davidson @thomasdavidson.bsky.social · 3d

There is one week left to apply to join us at Rutgers! We're hiring an Assistant Professor in Computational Sociology as part of a cluster of new hires in data science and AI.

Applications are due next Wednesday, 10/15.

Assistant Professor in Computational Sociology

The Department of Sociology at Rutgers University, New Brunswick, seeks applications for a tenure-track position at the Assistant Professor level specializing in Computational Sociology. The search i...

jobs.rutgers.edu

10 10

Maria Antoniak @mariaa.bsky.social · 3d

Another oral talk, this one by @stellali.bsky.social, discussing "PrefPalette: Personalized Preference Modeling with Latent Attributes."

Communities prefer different kinds of responses, which prioritize specific values. Aggregating preferences over communities would lose that signal.

#COLM2025

PrefPalette: Personalized Preference Modeling with Latent Attributes

Personalizing AI systems requires understanding not just what users prefer, but the reasons that underlie those preferences - yet current preference models typically treat human judgment as a black bo...

arxiv.org

1 12

Maria Antoniak @mariaa.bsky.social · 3d

At the end of the talk, he said he couldn't provide a full answer to that question but that we should each consider it. He said that he thinks there are potential upsides that he finds worth it, and that's why he's working on LMs. But he's not sure.

(paraphrasing a little bit from memory)

2 4

Maria Antoniak @mariaa.bsky.social · 3d

Now we're hearing a talk by @valentinhofmann.bsky.social about "Fluid Language Model Benchmarking."

Computerized adaptive testing is used for humans (like the GRE). This work adopts Item Response Theory from education to measure benchmark characteristics.

#COLM2025

Fluid Language Model Benchmarking

Language model (LM) benchmarking faces several challenges: comprehensive evaluations are costly, benchmarks often fail to measure the intended capabilities, and evaluation quality can degrade due to l...

arxiv.org

1 11

Maria Antoniak @mariaa.bsky.social · 3d

Question from the audience (didn't catch the name): Why are computer scientists the ones who should solve this problem? They lack expertise, and there are other people who have been studying these kinds of harms for a very long time.

#COLM2025

3 10

Maria Antoniak @mariaa.bsky.social · 3d

"What problems you're scared of depend on how good you think the LLMs will get"

"Please be willing to change your mind."

"This is COLM. We made the models, it's our job to fix it. How are you going to change your research agenda?"

#COLM2025

2 15

Maria Antoniak @mariaa.bsky.social · 3d

"I'm not arguing that everyone should work on all of these problems, I don't know how to work on this problem. But we should work on them scientifically and on a spectrum of problems."

Argues that these different kinds of risks are NOT just distractions from each other.

#COLM2025

2 9

Maria Antoniak @mariaa.bsky.social · 3d

Final risk is misalignment. (NB: You-know-who again. I'm not going to bother advertising.)

Nicholas expresses some skepticism but also asks that we don't immediately dismiss these risks.

"That's something that only happens in scifi... well then we live in scifi."

#COLM2025

1 3

Maria Antoniak @mariaa.bsky.social · 3d

Another example of misuse: dangerous capabilities (bio weapons), both OpenAI and Anthropic have expensive safeguards that run on every single query to look for these kinds of dangers

Are bio weapons the most important risk? Maybe, maybe not

#COLM2025

1 3

Maria Antoniak @mariaa.bsky.social · 3d

Another example of misuse: mass surveillance, "with language models you have the potential to watch everyone"

Quotes Larry Ellison on how "citizens will be on their best behavior" and notes that Oracle recently invested billions in OpenAI

#COLM2025

1 9

Maria Antoniak @mariaa.bsky.social · 3d

Misuse: Shows benchmark showing increasing success of models at identifying vulnerabilities, discusses ransomware at scale (they ran simulations on the Enron email dataset, Claude automatically found someone have an affair) and the first real example of this was recently discovered

#COLM2025

1 4

Maria Antoniak @mariaa.bsky.social · 3d

Now job replacement. "Things might be fine in 30 years but the next 20 years could be hard."

"Sometimes people claim that this worry is just something people say to sell stuff, but Dario has proposed taxing tokens (?) and offering to lose money is a sign people really believe this"

#COLM2025

1 6

Maria Antoniak @mariaa.bsky.social · 3d

Now discussing misinformation and Elon Musk.

"Previously no one person had so much power over information. If you control the language model, you have a significant amount of power about how people see the world."

#COLM2025

1 10