iiinova — where your digital self keeps evolving

Adina Williams @adinawilliams.bsky.social · 02/09/2026

Gonna be at COLM'26? Meta has a networking mixer on Oct 6! Check here for more info and to register: events.atmeta.com/metanetworkingmixercolm2026

events.atmeta.com

Meta Networking Mixer at CoLM 2026

Save the date! You won’t want to miss this event from Facebook.

020

Reposted by Adina Williams

Karen Ullrich (s/h) @karen-ullrich.bsky.social · 30/06/2026

We are launching a new blog; Reliable-AI.Review. First post is up: On the Impossibility of Mitigating AI Jailbreaks.

195

Adina Williams @adinawilliams.bsky.social · 10/06/2026

🪩Meta is hosting a networking event at ACL'26🪩 Learn more here: aclnetworkingevent2026.splashthat.com

aclnetworkingevent2026.splashthat.com

071

Adina Williams @adinawilliams.bsky.social · 28/05/2026

AI overview out there still reminding us of most frequent class baselines. I remember the not-so-distant days when everyone was aware of "2" bias in answers to "how many" questions (Agarwal et Al 2017 et subseq). I keep thinking maybe it's time to revisit these things.

arxiv.org

150

Adina Williams @adinawilliams.bsky.social · 03/05/2026

I had a fantastic time at the 2026 Harrington Symposium this week at UT Austin. It was wonderful to be able to dig into more science of AI with brilliant researchers across many specialties and viewpoints! Many things to think about! harrington.utexas.edu/faculty-fell...

harrington.utexas.edu

Harrington Faculty Fellows Symposium 2026

081

Reposted by Adina Williams

Noa Garcia @noagarciad.bsky.social · 26/04/2026

Today at 9am-5pm, we will host the first AI for Peace workshop at @iclr-conf.bsky.social With an excellent line-up of speakers: @pjreddie.bsky.social, @alexhanna.bsky.social, @davidthewid.bsky.social, Pier Luigi Dovesi, @giselalujan.bsky.social Come if you are worried about the use of AI for war.

1163

Reposted by Adina Williams

Sung Kim @sungkim.bsky.social · 14/04/2026

All elementary functions from a single binary operator He shows that a single binary operator, eml(x, y) = exp(x) − ln(y), together with the constant 1, generates the standard repertoire of a scientific calculator. Paper: arxiv.org/abs/2603.218...

4546

Adina Williams @adinawilliams.bsky.social · 25/03/2026

Thanks to @bayesianboy.bsky.social, Anya and everyone at Princeton Natural & Artificial Minds for the invitation!

180

Reposted by Adina Williams

Anna Rogers @annarogers.bsky.social · 03/12/2025

📢 Unhappy about peer review at NLP/ML conferences? Here's a chance to do smth about it, and get an interdisciplinary PhD! NLP+discrete optimization at @itu.dk Project: interpretable methods for paper-reviewer matching Apply by Jan 6 for position DC6: www.cords-dn.at Master degree needed

cords-dn.at

1227

Reposted by Adina Williams

Mark Ibrahim @markibrahim.bsky.social · 10/12/2025

Want to teach AI agents to use apps like humans? Get started with digital agents research using OpenApps, our new Python-based environment.

133

Adina Williams @adinawilliams.bsky.social · 06/12/2025

This is super cool! I had been contemplating starting a project like this a few years ago to actualize a vague interest I have in the math of knots... but this benchmark is probably much better than I could've managed. Can't wait to take a closer look 🪢!

191

Adina Williams @adinawilliams.bsky.social · 05/12/2025

We've create a simple and easy framework for documenting AI evaluation methods! Check it out and submit some for your own datasets with a PR! You can learn more here: github.com/facebookrese... And here: arxiv.org/abs/2512.04062

github.com

030

Adina Williams @adinawilliams.bsky.social · 01/12/2025

My spouse is doing a week of science communication (in German) this week. Topics may include cognitive science, linguistics, AI, political economy, science and technology studies, maybe even some philosophy too. Check it out👨‍🔬!

050

Reposted by Adina Williams

Tom McCoy @rtommccoy.bsky.social · 14/11/2025

🤖🧠I'll be considering applications for PhD students & postdocs to start at Yale in Fall 2026! If you are interested in the intersection of linguistics, cognitive science, & AI, I encourage you to apply! PhD link: rtmccoy.com/prospective_... Postdoc link: rtmccoy.com/prospective_...

Top: A syntax tree for the sentence "the doctor by the lawyer saw the artist".

Bottom: A continuous vector.

23713

Reposted by Adina Williams

Jennifer Hu @jennhu.bsky.social · 10/11/2025

New work to appear @ TACL! Language models (LMs) are remarkably good at generating novel well-formed sentences, leading to claims that they have mastered grammar. Yet they often assign higher probability to ungrammatical strings than to grammatical strings. How can both things be true? 🧵👇

Screenshot of a figure with two panels, labeled (a) and (b). The caption reads: "Figure 1: (a) Illustration of messages (left) and strings (right) in toy domain. Blue = grammatical strings. Red = ungrammatical strings. (b) Surprisal (negative log probability) assigned to toy strings by GPT-2."

29220

Reposted by Adina Williams

Kyle Mahowald @kmahowald.bsky.social · 07/10/2025

UT Austin Linguistics is hiring in computational linguistics! Asst or Assoc. We have a thriving group sites.utexas.edu/compling/ and a long proud history in the space. (For instance, fun fact, Jeff Elman was a UT Austin Linguistics Ph.D.) faculty.utexas.edu/career/170793 🤘

sites.utexas.edu

UT Austin Computational Linguistics Research Group – Humans processing computers processing humans processing language

14126

Reposted by Adina Williams

Hagen Blix @hagenblix.bsky.social · 16/09/2025

We wrote a thing about AI, fascism, and why framing this as "hype" is too apolitical www.liberalcurrents.com/deflating-hy...

liberalcurrents.com

Deflating “Hype” Won’t Save Us

The problem with AI isn’t hype. The problem is who and what it’s useful for.

412551

Adina Williams @adinawilliams.bsky.social · 15/07/2025

Our team is hiring a postdoc in (mechanistic) interpretability! The ideal candidate will have research experience in interpretability for text and/or image generation models and be excited about open science! Please consider applying or sharing with colleagues: metacareers.com/jobs/2223953961352324

careers.com

0115

Reposted by Adina Williams

Hagen Blix @hagenblix.bsky.social · 03/07/2025

Ingeborg and I wrote a thing about "hype", and why we think that framing AI through that lens is increasingly inadequate - check it out!

hagenblix.github.io

Deflating “Hype” Won’t Save Us

By Hagen Blix & Ingeborg Glimmer

22610

Reposted by Adina Williams

BlackboxNLP @blackboxnlp.bsky.social · 23/06/2025

Have you heard about this year's shared task? 📢 Mechanistic Interpretability (MI) is quickly advancing, but comparing methods remains a challenge. This year at #BlackboxNLP, we're introducing a shared task to rigorously evaluate MI methods in language models 🧵

1164

Adina Williams @adinawilliams.bsky.social · 12/06/2025

Excited to share the IntPhys 2.0 benchmark for intuitive physics understanding of video! It's super challenging for models but not humans, making it a great testbed for pushing modeling improvements. Check it out: ai.meta.com/research/pub...

ai.meta.com

IntPhys 2: Benchmarking Intuitive Physics Understanding In Complex Synthetic Environments | Research - AI at Meta

We present IntPhys 2, a video benchmark designed to evaluate the intuitive physics understanding of deep learning models. Building on the original...

020

Adina Williams @adinawilliams.bsky.social · 24/05/2025

Come by to our panel at APS to share your thoughts, and ask us all the hard stuff!

041

Adina Williams @adinawilliams.bsky.social · 23/05/2025

Check it out! Guy et al. explores the impact of format on function vectors, and invites further conversation about what it would mean to have universal goal representations in LLMs. (I've hung around interp communities for a while but this is my 1st mech-interp project. feedback much appreciated!)

0110

Adina Williams @adinawilliams.bsky.social · 20/05/2025

COLM 2025 workshops are announced now, check it out: colmweb.org/Workshops.html

colmweb.org

COLM 2025: Workshops

051

Adina Williams @adinawilliams.bsky.social · 23/04/2025

This is such a fun example of LM weirdness (which also shows how they match form over fact!) More linguistically: it looks like ending a query with "meaning" triggers the bot to accommodate the presupposition that the input contains an idiom! (Hard to run normal preposition tests here tho)

051

Adina Williams @adinawilliams.bsky.social · 11/04/2025

Happy to be at CUNY today to workshop about the theoretical foundations of AI! Fun to have several perspectives---math, philosophy, linguistics and more---together in one place! w/@giannig.bsky.social @seiller.bsky.social J. Terilla et al. itsatcuny.org/calendar/202...

itsatcuny.org

Foundations of AI — Initiative for the Theoretical Sciences

Friday, 11 April 9:30 AM - 4:00 PM Rooms 9206/9207, Graduate Center CUNY Join us at the Initiative for Theoretical Sciences for an interdisciplinary exploration of the foundations of artificial i...

071

Adina Williams @adinawilliams.bsky.social · 12/03/2025

Happy to share that the paper describing the AILuminate v1.0 benchmark is now out! arxiv.org/abs/2503.05731 The benchmark is designed with @mlcommons.org to assess LLM risk and reliability across 12 hazard categories. AILuminate is available for testing models and helping ensure safer deployment!

arxiv.org

AILuminate: Introducing v1.0 of the AI Risk and Reliability Benchmark from MLCommons

The rapid advancement and deployment of AI systems have created an urgent need for standard safety-evaluation frameworks. This paper introduces AILuminate v1.0, the first comprehensive industry-standa...

031

Reposted by Adina Williams

Hagen Blix @hagenblix.bsky.social · 11/03/2025

Our book "Why We Fear AI" is out today! Hopefully it can help make sense out of some of the terrifying stuff that's happening these days, and what AI and capitalism have to do with it! Get it directly from the publisher or wherever you get your books! www.commonnotions.org/why-we-fear-ai

commonnotions.org

Why We Fear AI — Common Notions Press

1132

Reposted by Adina Williams

Common Notions Press @commonnotions.bsky.social · 11/03/2025

WHY WE FEAR AI is out today! Industry insiders @hagenblix.bsky.social and Ingeborg Glimmer dive into the dark, twisted world of AI to demystify the many nightmares we have about it. One of the best ways to face your fear is to confront it—order a copy of WHY WE FEAR AI: buff.ly/1tWhkx8

Graphic featuring artwork from Why We Fear AI's cover. A black and white hand is reaching out toward text that says" We'll see how capitalism and class shape both the actual tools called ai and the ai nightmares."

063

Adina Williams @adinawilliams.bsky.social · 15/02/2025

My spouse co-wrote a book! It's about AI and what peoples' fears about it actually mean. Go check it out 👇

0120

Reposted by Adina Williams

Jean-Rémi King @jeanremiking.bsky.social · 07/02/2025

Two new studies from our team we're particularly happy about: Study 1: ai.meta.com/research/pub... Study 2: ai.meta.com/research/pub... Blog 3: ai.meta.com/blog/brain-a...

28221

Adina Williams @adinawilliams.bsky.social · 26/01/2025

We had two papers accepted at NAACL'25! 🥳 - Improving model evaluation using SMART filtering of benchmark datasets (Gupta et al. arxiv.org/pdf/2410.20245) - On the role of speech data in reducing toxicity detection bias (Bell et al. arxiv.org/pdf/2411.08135)

arxiv.org

0121

Adina Williams @adinawilliams.bsky.social · 17/01/2025

My spouse co-wrote a book about AI! Of course, I'm biased, but I think there's a TON of thought provoking stuff inside. Check it out/preorder 👇

070

Adina Williams @adinawilliams.bsky.social · 13/01/2025

Another post #LSA2025 thought: what's going on linguistically with "at scale" ("at a large enough scale for XYZ"). And maybe it's generalizing? "implement [AI] safely as they move at pace” www.theguardian.com/politics/2025/jan/12/mainlined-into-uks-veins-labour-announces-huge-public-rollout-of-ai

theguardian.com

‘Mainlined into UK’s veins’: Labour announces huge public rollout of AI

Plans to make UK world leader in AI sector include opening access to NHS and other public data

150

Adina Williams @adinawilliams.bsky.social · 11/01/2025

On the heels of #LSA2025 I read M. Tavakoli's simply fascinating piece on ventriloquism. An articulatory study of ventriloquist performance would be *so* cool! Feels like a goldmine for sociophon + maybe even wild new data on McGurk? Someone do it pls! Read here: www.nplusonemag.com/issue-49/ess...

1222

Adina Williams @adinawilliams.bsky.social · 09/01/2025

Excited to present in-process work on the syntactic complexity of LM outputs with Dexter Ju & @hagenblix.bsky.social at #LSA2025 in Philadelphia! Come on by Salon H in 30 min to chat about LMs and what they mean for cognitive science and linguistics! Learn more here: web.cvent.com/event/40d941...

web.cvent.com

LSA 2025 Annual Meeting.

0111

Reposted by Adina Williams

Conference on Language Modeling @colmweb.org · 18/12/2024

Announcement #2: our call for workshops is up! 💥 colmweb.org/cfw.html And excited to announce the first ever workshop chairs for COLM..... 🥁🥁🥁 COLM 2025 workshops will be led by Colin Raffel and @adinawilliams.bsky.social Workshop proposal deadline is April 14 🗓️

0114

Adina Williams @adinawilliams.bsky.social · 13/12/2024

Check it out! The reproducibility challenge is going physical! Hope to see some of you in Princeton in Aug '25

071

Adina Williams @adinawilliams.bsky.social · 11/12/2024

Our paper PRISM alignment won a best paper award at #neurips2024! All credits to @hannahrosekirk.bsky.social A.Whitefield, P.Röttger, A.M.Bean, K.Margatina, R.Mosquera-Gomez, J.Ciro, @maxbartolo.bsky.social H.He, B.Vidgen, S.Hale Catch Hannah tomorrow at neurips.cc/virtual/2024/poster/97804

blog.neurips

2679

Reposted by Adina Williams

MLRC 2025 - The ML Reproducibility Challenge @reproml.org · 10/12/2024

🎓 Checkout the #MLRC2023 posters at #NeurIPS 2024 this week: reproml.org/proceedings/ Also, important announcement for the next iteration coming this week! #MLSky #TMLR

reproml.org

Online Proceedings | MLRC

Machine Learning Reproducibility Challenge

021

Adina Williams @adinawilliams.bsky.social · 04/12/2024

So happy to see this benchmark launch! It's been a huge effort from so many contributors for months and months. Having broad-based coalition support for standardizing safety tools is an important step towards improving AI tools across the industry. Check it out!

020

Adina Williams @adinawilliams.bsky.social · 20/11/2024

#intro Looking forward to trying a new thing! Probably I'll post mostly about NLP and AI, with some linguistics and cognitive science thrown in for good measure.

0252