I did a QA with Quanta about interpretability and training dynamics! I got to talk about a bunch of research hobby horses and how I got into them.

🚨 Are you looking for a PhD in #NLProc dealing with #LLMs? 🎉 Good news: I am hiring! 🎉 The position is part of the “Contested Climate Futures" project. 🌱🌍 You will focus on developing next-generation AI methods🤖 to analyze climate-related concepts in content—including texts, images, and videos.

Madeleine I. G. Daepp, Alejandro Cuevas, Robert Osazuwa Ness, Vickie Yu-Ping Wang, Bharat Kumar Nayak, Dibyendu Mishra, Ti-Chung Cheng, Shaily Desai, Joyojeet Pal: Generative Propaganda https://arxiv.org/abs/2509.19147 https://arxiv.org/pdf/2509.19147 https://arxiv.org/html/2509.19147

Nishant Balepur, Matthew Shu, Yoo Yeon Sung, Seraphina Goldfarb-Tarrant, Shi Feng, Fumeng Yang, Rachel Rudinger, Jordan Lee Boyd-Graber A Good Plan is Hard to Find: Aligning Models with Preferences is Misaligned with What Helps Users https://arxiv.org/abs/2509.18632

I think this miiight be the best feed for machine-learning news on Bluesky. It strikes a good balance between new/popular and whitelist/regex factors in the algorithm.

Marji Gesick, the hardest single day bike race in America, was this past weekend, and for the first time, women racers buckled (finished in under 12 hours). One of those women, Kait Boyle, wrote a fantastic blog about the whole thing: kaitboyle.substack.com/p/unfinished... #bikesky

Deadline extended: Are you the new postdoc who will work with me in this great project on bidirectional alignment for citizen-facing AI solutions in an emergency triage context? Don't forget to apply by the 1st October! candidate.hr-manager.net/ApplicationI...

Also about AI, data, early days at Google, studying literature, and more. Worth a read!

Mian Zhong, Pristina Wang, Anjalie Field HICode: Hierarchical Inductive Coding with LLMs https://arxiv.org/abs/2509.17946

I’m in Copenhagen today and tomorrow and Paris the rest of the week (for the Humanities and Artificial Intelligence Virtual Institute hosted by Schmidt Sciences). Say hi if you’re around! 👋

Kultura starts tomorrow 🎨 8 weeks exploring Ukraine’s visual culture — from medieval Kyiv to contemporary art & feminism. Join full course or individual lectures online. 📅 22 Sept – 10 Nov 👉 Sign up: tickettailor.com/events/ukrainianinstitutelondon/1840178

Some more info in this WaPo story. The crux is that she told the bot repeatedly about her plans, and the company did nothing. But there is a lot going on in all these stories about designed-to-be-addictive tech, interweaving of offline and chatbot and online forum and fictional worlds…

wearing my “world without caesars” shirt today for the special occasion of going through TSA (and everything else that happened this week) but i fear it is too subtle and i need a better shirt

Created a small tool for updating a Slack channel when new papers are added to a Zotero library. Totally free using Google Cloud Run's basic tier. levon003.github.io/2025/09/19/z...

I wish students understood in most empirical AI research there’s a huge scientific advantage from being constitutionally excited by math vs intimidated, but very little additional gain from being actually “good” at math. Maybe they’d be less intimidated if they didn’t feel they had to be “good”.

Very excited that my paper with @katakeith.bsky.social is now out in @polanalysis.bsky.social. We investigate whether LLMs actually follow the instructions/definitions provided in codebooks, propose some diagnostics, and release a new evaluation dataset. www.cambridge.org/core/journal...

#computationalsocialscience Our work on using VLLMs for semantic visual clustering is out on SSCR: journals.sagepub.com/doi/10.1177/... There is a lot happening in the space of computational analysis of visual content as we acknowledge more and more the visual nature of contemporary communication.

excited to share that we are following through on our earlier commitments and putting together an independent+neutral organization to house the DID PLC system, includes the directory service

I explore these ideas and more in my preprint, “Hallucinating with AI: AI psychosis and distributed delusions”, where I argue that we need to move beyond thinking about AI “hallucinating at us” to understanding how we “hallucinate with AI” #philsky arxiv.org/abs/2508.19588

The real existential threat isn't AI — it's billionaires. My latest, for @theatlantic.com: www.theatlantic.com/books/archiv...

Trying to keep my professional chill but I’m SO excited Carnegie Mellon is launching a cluster hire in computational humanities—MULTIPLE JOBS! 1. Asst Teaching Track Prof in Computational Humanities - apply.interfolio.com/173622 2. Asst Tenure Track Prof in CH - apply.interfolio.com/173626

🇺🇸 Let’s start a new nation, founded on the concept of free speech. We can conceive it based on liberty, and dedicate it to the proposition that all people are created equal. Too radical?

Given funding uncertainty, we wrote about why depts & faculty should try to communicate openly with PhD applicants about any changes to their admissions processes, including if it's "nothing has changed". Let's do what we can (when we can) to prevent cascading anxiety. cra.org/crn/2025/09/...

TIL that the CU Boulder campus uses more water for irrigation each year than Seattle receives in rainfall in the same time span.

My first Princeton AND my first systems paper! In this work, we develop a system (Bonsai 🌱) to enable users to design intentional social media feeds. Our findings identify trade-offs between engagement- and intention-driven social feeds. Can we build towards hybrid approaches in the future?

🚨Microsoft Research NYC is hiring🚨 We're hiring postdocs and senior researchers in AI/ML broadly, and in specific areas like test-time scaling and science of DL. Postdoc applications due Oct 22, 2025. Senior researcher applications considered on a rolling basis. Links to apply: aka.ms/msrnyc-jobs

“Keep your head down, follow without questioning, adjust field of study” It’s alarming how many academics are already lining up to join this camp and publicly recruiting others to do the same, including US citizens and profs with tenure.

📢 #CSCW is experimenting with a new format: rolling submissions 🎉 Submit your work when it’s ready! ✔️ PACM publication + CSCW presentation ✔️ Healthier work-life balance ✔️ Faster reviews and more opportunities for revisions More info: medium.com/acm-cscw/csc... #HCI

If you’re like me and feeling increasingly hopeless and powerless about rising fascism, here is something you can do about it ⬇️

This is the experiment we're trying next: reciprocal.reviews. Earn tokens for reviews by writing adequate quality reviews, spend them on submissions. Earn the privilege to publish. We're at 70% on the beta, and should launch an ACM TOCE pilot next year some time.

Are there conceptual directions in VLMs that transcend modality? Check out our COLM oral spotlight 🔦 paper! We use SAEs to analyze the multimodality of linear concepts in VLMs with @chloesu07.bsky.social, @thomasfel.bsky.social, @shamkakade.bsky.social and Stephanie Gil arxiv.org/abs/2504.11695

This came up in my class yesterday. We read @oms279.bsky.social’s wonderful paper “From Codebooks to Promptbooks” (highly recommend for teaching), but got stuck on the issue of non-random errors, which is a not a unique issue to LMs. Traditional methods… Paper: journals.sagepub.com/doi/10.1177/...

This comes from a recent talk I gave at the SkAI institute conference titled: "Structured Creativity in Science: AI for Hypothesis Generation and Research Ideation." Learn more here: drive.google.com/file/d/1_dLd...

📢 New #COLM2025 paper 📢 Standard benchmarks give every LLM the same questions. This is like testing 5th graders and college seniors with *one* exam! 🥴 Meet Fluid Benchmarking, a capability-adaptive eval method delivering lower variance, higher validity, and reduced cost. 🧵

Our tool Riveter💪 used for a creative and interesting study of fan fiction! Riveter helps you work with “connotation frames” (verb lexica) to measure biases in your dataset. @julianeugarten.bsky.social’s overview and explanations are really clear, highly recommend! github.com/maartensap/r...

This study not only asks human experts to score AI- and human-generated ideas, but also to *execute* those ideas (a nice setup!) AI-generated research ideas *sounded* better (high estimated effectiveness score) but their score drops below human ideas after the execution. arxiv.org/abs/2506.20803

Modeling Culture explores AI not only as a technology to critique, but as a new mode of inquiry for research and interpretation. Three interconnected components will unfold in 2025-26: a faculty and grad seminar, a public lecture series, and an open-access curriculum. Public talk details in thread👇

UMSI is running multiple searches this year, starting with the John Derby Evans Professor in Information, at the Assistant or Associate level! This is open to anyone working at the intersection of tech and society, with a closing date of Nov 1, 2025. Please share! www.si.umich.edu/people/facul...

Colorado updates: - I’ve developed a fear of the sun, it’s unreasonably strong and far too close - I wore a neon jacket while biking - I’ve eaten way too many burritos - I carefully studied the Telluride Film Festival paparazzi photos for mountain style inspo - I signed up for mobile fire alerts

Hi Friends! My coauthors @tedunderwood.me @gworthey.bsky.social @profdownie.bsky.social and I are excited to share our new article in DSH @adho-org.bsky.social @eadhorg.bsky.social "Comparative analysis of classics book review data created by users across Douban and Goodreads". Link in comment:)

“AI is a narrative” — yes! And other narratives are possible.

"The report identifies nascent but powerful CoP that already produce nuanced criticism of the Big Tech-driven AI development ecosystem, while they also imagine, develop, and, at times, deploy AI systems informed and guided by commitments to decoloniality, feminism, antiracism, and post-capitalism."

I’ve been lucky to get the chance to work with some incredibly creative and talented undergrad researchers through this program! Help us spread the word to students who might be interested!

Does anyone know of work on what the use of fiction in studying psychology tells us about what people learn from narratives in general? Empirical reading studies or psycholinguistics folks? Many psych experiments aren't live-action role playing (e.g., Stanford prison experiment) but reading stories.

I’m in Atlanta today for a workshop on “AI. Culture. Democracy. Now. How the humanities can lead in a time of crisis.” Looking for some rays of hope and a sense of purpose in these strange times!

Q. Who aligns the aligners? A. alignmentalignment.ai Today I’m humbled to announce an epoch-defining event: the launch of the 𝗖𝗲𝗻𝘁𝗲𝗿 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗔𝗹𝗶𝗴𝗻𝗺𝗲𝗻𝘁 𝗼𝗳 𝗔𝗜 𝗔𝗹𝗶𝗴𝗻𝗺𝗲𝗻𝘁 𝗖𝗲𝗻𝘁𝗲𝗿𝘀.

@loriemerson.net’s reading and performance were inspiring! Other networks not only are possible, they exist!

Important new article form my colleague Paula Westenberger on Artificial intelligence for cultural heritage research, and the challenges in UK copyright law and policy. www.frontierspartnerships.org/journals/eur...

We are so, so fortunate to have the ACL Anthology.