Hi everyone I was denied entry, detained, and deported from the USA over the last 48 hours because of my reporting on the Columbia student protests I arrived back in Melbourne hours ago and had my phone handed back to me upon landing

Stanford scholars introduced an open-source AI agent that learns how to navigate websites by mimicking childhood learning – an approach that could lead to more efficient, transparent, and privacy-conscious AI: hai.stanford.edu/news/an-open... @chrmanning.bsky.social @shikharmurty.bsky.social

Science is under threat in the US. @elife.bsky.social have commissioned a series of articles discussing the implications and what we can do. The first three articles are now live. More to follow: elifesciences.org/articles/106... elifesciences.org/articles/106... elifesciences.org/articles/106...

US tech firms - IME - rely heavily on grads of higher ed in general, and US higher ed in particular... And yet I've not heard much noise from the Pichais, Nadellas, Zuckerbergs, Cooks, Benioffs, Sus, Jassys, Huangs, etc. about attacks on higher education... Am I missing it or does it not exist?

1. @alisongopnik.bsky.social, Cosma Shalizi, James Evans and myself have a new piece in Science on "AI" Large Models, pushing back against much of the collective wisdom about what they can and can't do. Official below, unpaywalled at henryfarrell.net/large-ai-mod... . So why this now?

The New York Times is the first to put out comprehensive estimates on the cost of a year without U.S.A.I.D. and they’re higher than I thought: - 1.65 million deaths from AIDS - 500,000 from lack of vaccines - 550,000 from lack of food aid - 290,000 from malaria - 310,000 from TB

If you asked me to come up with ways to sabotage America from within, I wouldn’t do as good a job as Republicans have done this year. Just a bottomless pit of economic self owns.

The US 🇺🇸 under Trump is pretty decisively committed to unilateral on-whim trade tariffs. For countries like Canada 🇨🇦 and Australia 🇦🇺 that still believe in rules-based free trade, maybe it’s time to think much more seriously about getting the EU and then … moderndiplomacy.eu/2025/03/14/e...

I’m fascinated by both Schumer’s logics and by the reaction to his logics from the Dems. The former is Schumer’s failure to understand the broader anger toward Congress. The latter is the public’s failure to understand the trap that the GOP set re the administrative state. There is no win here.

An introductory talk by @chrmanning.bsky.social on “Large Language Models in 2025 – How much understanding and intelligence?” at the Workshop on a Public AI Assistant to Worldwide Knowledge at Stanford, covering 3 eras of LLMs, RAG, Agents, DeepSeek-R1, using LLMs, …. Video: youtu.be/5Aer7MUSuSU

I am concerned about AI but late at night, alone working on a proposal, I was glad ChatGPT had my back as I hit submit 😀.. Reminded me of @chrmanning.bsky.social’s mention in a talk of the 'Real World Utility Test' - early adoption of tech moves forward when it’s genuinely useful, concerns and all.

As @NIST Scientist Emeritus Ellen Voorhees said in 2000 (!), IR focused on “retrieving a ranked list of documents” but often “a user has a specific question and would much prefer that the system return the answer itself”. It’s super to see this becoming reality with neural LLMs!

In 2013, at AKBC 2013 and other workshops, I gave a talk titled “Texts are Knowledge”. This was well before there were any transformer LLMs—indeed before the invention of attention—and my early neural NLP ideas were rudimentary. 🔮 Nevertheless, the talk was quite prophetic!

The new Computing and Data Science (CoDa) building at @stanforduniversity.bsky.social is beautiful, with many lovely spaces, and a great, if already too crowded, Voyager Coffee coffee shop. news.stanford.edu/stories/2025...

In this collection of perspectives, @stanfordhai.bsky.social senior fellows offer a multidisciplinary discussion of what #DeepSeek means for AI and society at large. hai.stanford.edu/news/how-dis... @chrmanning.bsky.social @jlanday.bsky.social @yejinchoinka.bsky.social @amyzegart.bsky.social

How can it be that the Apple iOS data detector pop-up gives you options to call or copy but not text? We’re not in the twentieth century any more. The twenty-first century is almost a quarter done.

1. Today the NIH director issued a new directive slashing overhead rates to 15%. I want to provide some context on what that means and why it matters. grants.nih.gov/grants/guide...

Want to make a browser agent for *any* domain like banking or healthcare? We propose methods for training LLMs with open-ended, unsupervised interaction on live websites: ✅ OSS SoTA on WebVoyager ✅ world's smallest high-performing web-agent Try it here: nnetnav.dev

Maybe vilifying China & Chinese people, and increasing visa hassles & investigations aren’t smart moves for keeping the US lead in AI—given the number of engineers trained in China vs US? “Many Chinese students are not that interested in full-time jobs in the US.” restofworld.org/2025/china-a...

I went to a Silicon Valley AI retreat in 2019. Turned out it was run by EA and Longtermists before I knew what those things were. One activity was to envision future scenarios involving AI. I told a story about the dangers of regulatory capture. I was shunned for the rest of the event.

“Those who don’t know history are destined to repeat it.” A good historical perspective: www.nytimes.com/2025/01/23/o...

Got my first spam message on @signal.org I guess that’s evidence that the platform is growing?

New developments on our 'LLMs can do metalinguistics' paper! We created sentences and had GPT3.5, 4, o1 and Llama 3 analyze them as a linguist would. We had 3 graduate students in linguistics evaluate how good linguists LLMs really are.

I feel like I can see America changing before my eyes* “The only lasting truth Is Change. God Is Change.” – Octavia Butler, Parable of the Sower *Okay, maybe this feeling was exaggerated this morning by the Navy vs. Oklahoma Armed Forces Bowl game being on in the barbershop, but still …

Six months ago someone put a for-loop around GPT-4o and got 50% on the ARC-AGI test set and 72% on a held-out training set redwoodresearch.substack.com/p/getting-50... Just sample 8000 times with beam search. o3 is probably a more principled search technique...

Just got back from Vancouver and I was amazed at how their downtown is full of cool little small businesses. The secret is … socialism? The government takes over leases of long-vacant spaces and sublets small subdivisions to locally-owned businesses and nonprofits. www.postalley.org/2020/07/20/h...

This is so far our best call for papers! We completely re-thought what a CFP is, and.... mostly copied last year's while surgically increasing some numbers

At #NeurIPS2024, the keynote speaker perpetuated explicit racist stereotypes against Chinese students. Generalizations against a community subject to discrimination, even as an “example”, further provokes discrimination. Below, video of an audience member’s perfect response.

Come visit our poster "MoEUT: Mixture-of-Experts Universal Transformers" on Friday at 4:30 in East Exhibit Hall A-C #1907 on #NeurIPS2024. With Kazuki Irie, Jürgen Schmidhuber, Christopher Potts and @chrmanning.bsky.social.

Thanks to @alexpasternak.bsky.social and @fastcompany.com for this lovely interview and review of The AI Mirror, part of their AI20 (with @abeba.bsky.social, @sashamtl.bsky.social + others)

Absolutely. I worry about an entire generation that can’t go from a blank page to a coherent, logical, articulate argument on their own. This process is not just writing: it is thinking. If we cede all our writing to AI we cede also our very ability to think.

20+ years late, I see Gosse Bouma wrote a lovely review of Avery Andrews and my Complex Predicates and Information Spreading in LFG in Studies in Language: “this book makes a valuable contribution to the theory of LFG and feature-based theories of grammar” www.jbe-platform.com/content/jour...

The extraordinary recent takeover of ML/AI by #NLP is well-known but insufficiently reflected on. Look at the @neuripsconf.bsky.social tutorials in 2024! neurips.cc/virtual/2024... 14 tutorials; 6 have "LLM" in the title; 4 more cover foundation models, with large NLP coverage. That's > 70% 😲

Natural Language Processing—artificial intelligence that uses human language—has been on a roll lately. You’ve probably noticed! So the Stanford NLP Group has been growing, and diversifying into lots of new topics, including agents, language model programs, and socially aware #NLP. nlp.stanford.edu

America has long been a magnet for talent from around the world, powering an amazing innovation ecosystem. But we're working hard to destroy that advantage. 🤷‍♂️

I had the incredible good fortune to be part of the team that created #GivingTuesday in 2012. Tomorrow, December 3, is #GivingTuesday, a day for the simple and uncomplicated practice of generosity.

I would put this even more strongly: open source AI is probably our only realistic chance to avoid a terrifying increase in concentration of power. I do not want to live in a world where the people with all the money also have all the intellectual power.

I think we saw this in the AI/ML sphere, not just in politics …

ACL syntax track reviewers >> almost any other conference. These folks care about their sub-field and i learn something new every time!

The authors of ColPali trained a retrieval model based on SmolVLM 🤠 TLDR; - ColSmolVLM performs better than ColPali and DSE-Qwen2 on all English tasks - ColSmolVLM is more memory efficient than ColQwen2 💗 Find the model here huggingface.co/vidore/colsm...

I take no joy in admitting that the new Claude really does produce conversational outputs I would call insightful if a person said them

🚨 NeurIPS 2024 Spotlight Did you know we lack standards for AI benchmarks, despite their role in tracking progress, comparing models, and shaping policy? 🤯 Enter BetterBench–our framework with 46 criteria to assess benchmark quality: betterbench.stanford.edu 1/x

New York and California ought to pass some voting reforms that include both rapid counting on par with Florida’s, but also compulsory voting. @kjephd.bsky.social had already persuaded me of the importance of the latter. The depressed blue state turnout this time around just highlighted it again.

📄👀: Consonant lengthening marks the beginning of words across a diverse sample of languages www.nature.com/articles/s41...

A great list in general, but cuffed to be first! Thx, @jacobeisenstein.bsky.social !

easy first #booksky #nlp

By demand, I've created the final starter pack in my ML Personality Starter Pack Series. I'm uncertain who belongs in this starter pack and so if you think you better fit in the Grumpy ML or Unreasonably Upbeat ML starter packs, let me know. (Self) nominations welcome go.bsky.app/5Suyk58

#EMNLP2024 was a fun time to reconnect with old friends and meet new ones! Reflecting on the conference program and in-person discussions, I believe we're seeing the "Google Moment" to #IR research play out in #NLProc. 1/n

It's a bit strange to me how passively users are discussed and modeled in misinformation research. As in, yes, there are various contexts under which an information ecosystem is imposed upon you but, in most cases, people make pretty conscious *choices* on what to consume & how..