A ton of attention over the years goes to plots comparing open to closed models. The real trend that matters for AI impacts on society is the gap between closed frontier models and local consumer models. Local models passing major milestones will have major repercussions. buff.ly/ccMJydQ

Claude, "We all know among Sauron's many evils was that he ran Mordor using an Excel spreadsheet with multiple tabs. Show me the spreadsheet" It made 12 tabs "so bureaucratically complex that even the Eye of Sauron would need reading glasses to review it." Some very funny stuff. Creative, even.

Gemini Live is giving me the confidence to speak my second language The anxiety I experience when trying to speak my second language disappears when I use Gemini Live. #ai #gemini #geminiai

This is a cool paper that suggests that AI agents can indeed be used for social science experiments, but that just using a chatbot isn't good enough, instead prompts developed based on social & game theory makes AI agent actions predictive of real human outcomes. benjaminmanning.io/files/optimi...

Duolingo’s recent earnings shows the gap between online vocal AI critics and businesses real-world value. Despite the backlash, Duolingo’s revenue and daily users jumped 40%. The AI hype is wild, but so is the value people are getting. This isn’t blockchain, it’s more like mobile or the internet

"In collaboration with the @archive.org TV News Archive we examined 14,749 evening news broadcasts spanning the big three networks from July 2010 to present, asking Gemini 2.5 Flash Thinking to make an index of the stories covered in each broadcast, geography, sentiment, frame, narrative structure.

We may have the chance to hire an outstanding researcher 3+ years post PhD to join Tarleton Gillespie, Mary Gray and me in Cambridge MA bringing critical sociotechnical perspectives to bear on new technologies. jobs.careers.microsoft.com/global/en/jo...

🤯

🥇 I am so happy to see the Google DeepMind Gemini team **OFFICIALLY** achieved gold-medal performance. Not only did we get 5 out of 6 questions correct. What's more amazing is the official IMO graders find those answers to be clear, precise and easy to follow! 1/ deepmind.google/discover/blo...

"[video game] as a community theater production" may be one of the most delightful Veo 3 Fast prompts Please enjoy, in order: GTA, Pokemon, Mario Kart, The Witcher 3, Stardew Valley, Tetris, Mortal Kombat, The Sims, & Death Stranding. The whole prompt was the one above. And the glitches add to it

If you're at @icmlconf.bsky.social this week, come check out our poster on "Position: Evaluating Generative AI Systems Is a Social Science Measurement Challenge" presented by the amazing @afedercooper.bsky.social from 11:30am--1:30pm PDT on Weds!!! icml.cc/virtual/2025...

The #FAccT and #ICML tags are good places to find some brilliant folks

The best article ive read so far on social impacts: henryfarrell.net/large-ai-mod...

Did a starter pack a while back too of scholars and practitioners who cover this angle in case that’s useful go.bsky.app/5sFqVNS

If it's helpful, I followed the recommendations in this post from @nsaphra.bsky.social and as a result have a much more interesting and ML-focused feed: nsaphra.net/post/bsky/

These are the ML feeds I've pinned that now get mixed into my main feed: - ML Feed: Trending bsky.app/profile/smcg... - MLSky: bsky.app/profile/alex... - GenAI Bluesky Network: bsky.app/profile/hell... - Nuanced AI Commentary: bsky.app/profile/dame...

I can recommend following: @simonwillison.net, @metr.org, @abeba.bsky.social, @mmitchell.bsky.social , @wang.social, @sashamtl.bsky.social, @justinhendrix.bsky.social, @bvlsingler.bsky.social Some of the people I follow to keep up with that side of things. :)

I would love to see more discussion on ML/AI on bsky. Particularly more nuanced discussions of the societal impacts. If you tag in the relevant post, I'll be sure to repost it!

Still not a lot of ML talk on bsky (at least in my feed), hence paper Sunday: my two most interesting recent reads - H Nets arxiv.org/abs/2507.07955 - Energy Based Transformers arxiv.org/abs/2507.02092

I’ve loved everything about the Red Lobster turnaround: • A young Nigerian immigrant CEO who previously ran PF Chang (which I love) • Listened to customers and made swift changes • Stars in ads, bringing real authenticity to the “we’re fixing it” vibe

Reasons you should donate to EFF this month: 1. We're turning 35! Wish us a happy birthday 🥳 2. We've got cool new swag--our gift to you 👕 3. We're advocating for your rights to privacy and free speech online 👩‍💻

Anyone at #facct2025 want to go to the acropolis this afternoon?

Claude Code, meet your rival: Gemini CLI 🔥🚀 Google just unleashed Gemini 2.5 Pro at the command line—free, open-source (Apache 2.0) , 1-million-token brain baked right into your terminal. Chat, code, search, script—all with zero tab fatigue.

The real Bear Grylls

The two most common questions I get asked about AI are “which AI should I use” and “how do I start using AI?” I wrote a short guide attempting to answer both questions. www.oneusefulthing.org/p/using-ai-r...

Current Polymarket probabilities: -Khamenei out as Supreme Leader of Iran in 2025? 63% -Will the Iranian regime fall in 2025? 29% -Iranian coup attempt before July? 12% -Iran Nuke in 2025? 17% -Will Iran close the Strait of Hormuz in 2025? 52% -US-Iran nuclear deal in 2025? 36%

veo 3: "three toy ships, one made of iron, the other of wood, and one out of loosely packed sugar, are dropped into a pool of water" AI video tools really do seem to be able to simulate physics well (but not perfectly) without having an underlying physics engine. Signs of a world model?

"GBH News could not find any other examples of the Harvard Corporation stripping a tenured professor of their status." Yes, because in other fraud cases I'm familiar with, the professors eventually resigned "voluntarily", sometimes after negotiations may have led to a golden parachute.

This is really disappointing to hear. I really enjoy Ethan's posts and perspective on AI.

You can just take academic papers and paste them into Gemini 2.5/ChatGPT o3/Claude 4 with the prompt "build me a game based on this paper, make it interesting and thematic but still conveying key findings" and get a tiny working educational game. (In this case, I used Gemini)

It's been 23 years since Bernstein's "10 tips on Writing the Living Web," an iconic guide to blogging. In 2025 I suspect most are no longer viable & some might never have worked for everyone. Three that still apply are: - write for a reason - write often - write tight alistapart.com/article/writ...

Next week is Google IO and there are many cool things happening! 🎉🎉🎉 One of them is this presentation I'll give with my college Omar talking about Gemma models It will be recorded and streamed! io.google/2025/explore...

Why We Think by Lilian Weng Test time compute and Chain-of-thought (CoT) have led to significant improvements in model performance, while raising many research questions. This post aims to review recent developments in how to effectively use test-time compute (i.e. “thinking time”) and why it helps

NEW ARTICLE: Want to use LLMs to extract information at scale in sociology? @eollion.bsky.social @oms279.bsky.social and C. Ton have you covered. Read "From Codebooks to Promptbooks," now in Sociological Methods & Research doi.org/10.1177/0049... (Preprint @socarxiv.bsky.social: osf.io/wjvfq_v1)

Slop farmer...is now a thing?

Sycophancy and the art of the model. Borderline self-indulgant to get to write about the messiness of RLHF to this extent, but simultaneously essential. A must read. GPT-4o-simp is a sign of things to come. www.interconnects.ai/p/sycophancy...

I find it totally crazy that the IRB didn't find the lack of informed consent problematic? That's one of the most important requirements of human behavioral research. I remember having to take multiple trainings on it during grad school to make sure that, no really, I understood how important it was

👀 papers.ssrn.com/sol3/papers....

Best airport food in the US

Google Gemini team loves this chart for some reason... aider.chat/docs/leaderb...

Keep the faith. Take action. Be honest about risk. Be prudent (with words, acquaintances, and technology). @juliaangwin.com and @amifieldsmeyer.bsky.social distill the wisdom of elders and of the ages into a primer for The Opposition™️ Your weekend #mustread

The Computing Research Association (@cra-hq.bsky.social) has a new trustworthy AI research fellowship. Intended for folks 1-3 years out of their PhD and with interdisciplinary experience in social sciences. cra.org/cra-trustwor...

This is a great reminder of the work that is needed to improve the science of AI evaluations.

What’s next in STS? Apply here to make the case:

It's a real shame that ABC shut down 538 right now...

Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems (264 pages) This survey provides a comprehensive overview, framing intelligent agents within a modular, brain-inspired architecture

OpenAI: "Users have told us that understanding how the model reasons ... helps build trust in its answers." Anthropic: "Do reasoning models accurately verbalize their reasoning? Our new paper shows they don't." www.anthropic.com/research/rea...

Extra desperately searching for a single emergency reviewer to step in for another emergency reviewer who has gone missing. Topic is safety/ethics and LLMs.