I defended my PhD 10 years ago today. That was the least remarkable thing that happened that day. Sharing something I wrote about it last year. Since then "the horrors persist but so do we." And with dignity.

These almost all are pretty optimistic on getting further below the error correcting threshold. Considering this was only achieved recently, it’s not clear how achievable this is.

Neutral atoms are exciting, but still lots of challenges. Hard not to be jaded after many cycles of hype. But all these works getting space requirements down are exciting, it seems almost all implementations struggle most with scaling qubit number.

The whole tech CEO celebrity culture the press created in the early 2000s has been such a disaster

Google Flips Its Fiber To PE.

I'm glad someone is calling out this trend. It's particularly pernicious in tech.

Here is a picture of a baby otter for no good reason.

Sandra Wong, the great granddaughter of Wong Kim Ark, is preparing to march at the #NoKings Rally in San Francisco. Today is the 128th Anniversary of the SCOTUS decision affirming birthright citizenship.

Discworld QOTD, from Feet of Clay

Regehr’s latest blog post is fantastic and makes me sad that the term “Useful Idiots” already has a well known meeting outside of tech

I’m at the Harvard Kennedy School today to share my own special brand of AI skepticism, and I’m glad I made the trip. I’ve heard more AGI-pilled, ahistorical, technosolutionist bullshit in the past three hours than I have all year. (This is a tribute to the select company I normally keep.)

I had not imagined this would be an issue for a conference. Great analysis by @peterhenderson.bsky.social

Omg??????? www.nytimes.com/2026/03/26/b...

Tasking an algorithm with making determinations about medical care can create unwarranted—and even discriminatory—delays or denials of necessary treatment. We're suing to get answers about Medicare's AI experiment. eff.org/press/relea...

how diffie hellman key exchange works (with as little math as possible)

Maybe LLMs will eat software development. But maybe they'll make software development more fun. RTWT: How to make LLMs more reliable at generating code by not giving them degrees of freedom to write bad code, by using tools we already have for writing good code. john.regehr.org/writing/zero...

There's a truly tragic version of Goodhart's law lurking in this article - "1000 targeting decisions per hour" meets stale data, leading to 100s of dead schoolgirls

Trump's newly announced council of scientific advisors is a startling departure from precious PCASTs: Instead of academic researchers or industry scientists, nine out of 13 members are billionaires, who have a combined wealth of over $900 billion. My reporting:

A truly wonderful article to unpack the nonsense discourse around the role of LLMs in the war machine.

I’ve had a longstanding policy of always doing whatever @friedler.net says and it hasn’t led be astray yet. So goes for these recommendations. CC: Congress

Excellent debunk of the Shy Girl AI allegations. www.tiktok.com/t/ZTk8eGAT1/

Very pleased to share our thoughts on AI & labor. @friedler.net, Andrew Schrank, Sue Helper, and I put forth an optimistic vision for the future of AI & work; we argue for a worker-first response to AI. www.brookings.edu/articles/a-p...

On rigor in mathematics. search.app/Y3h79

BREAKING: A jury in Los Angeles finds Instagram and YouTube liable in a landmark social media addiction trial.

Update ‼️ We are extending these #FAccT2026 deadlines for craft and tutorial sessions to Friday, March 27!

This happened to me earlier this year. Fortunately, the video creator emailed me and gave me the opportunity to opt out. But it definitely freaked me out, since the participants didn't consent to being filmed and published publicly.

This is quite the story.

Wow a big hit for AI2 and for public interest, open source AI research

look at this baby beaver and feel better briefly (@ wildheartrescue on TT)

I'm looking forward to visiting Carnegie Mellon University this week. I'll be giving a talk on 3/26 at noon at the AI Institute for Social Decision Making. For people in Pittsburgh, here's the link: events.cmu.edu/ai-sdm/event... I don't know if there will be an option for virtual streaming.

No, @washingtonpost.com Editorial Board, I remember when conservatives **lied** that the board would be a Ministry of Truth, when its founding documents prove that it was set up to *protect* fundamental rights while responding to disinfo. Glad I cancelled my subscription months ago.

Vietnam is Southeast Asia’s first country with a comprehensive AI law. Modeled on the EU’s risk-based approach, it emphasizes human oversight, risk classification, and digital sovereignty as part of Hanoi’s 2045 development push, reports Lam Le.

I spoke with @rollingstone.com about the Trump administration's new AI framework. At a moment when a majority of Americans are asking for AI guardrails, this proposal moves in the opposite direction. Rather that constrain companies, it releases them from accountability. 🧵 www.msn.com/en-us/news/p...

@alondra.bsky.social talks about the new AI policy doc from the administration. www.rollingstone.com/culture/cult...

The deadline for Brown AI Policy Summer School is coming up soon: March 27!

Emily Tucker, the Executive Director of @georgetownprivacy.bsky.social, remains stellar and GOATed. medium.com/center-on-pr...

My son found this lying around in his dorm in the throw pile and rescued it for me. #iykyk. Cc @cscheid.net

Chuck Norris challenged Death to a karate fight ... and alas, Death won. RIP. #chucknorrisfacts Chuck Norris, prolific action star and martial arts champion, dies aged 86 | Movies | The Guardian share.google/1N3jKA6vAf4Z...

Perfect

Real or slop? Can you tell which PL paper is real and which is slop? This is brilliant. slop.zackg.me

Congratulations to Charles H. Bennett and Gilles Brassard on receiving the 2025 ACM A.M. Turing Award! They are recognized for their essential role in establishing the foundations of quantum information science and transforming secure communication and computing. awards.acm.org/turing @umontreal

Quite the graph.

Not to toot our own horns but I feel like we packed so much in this one hour panel with @karenhao.bsky.social and John Palfrey. You can find the recording at www.youtube.com/live/I1tJnM8...

The new BBC podcast The interface with @karenhao.bsky.social raised this spectre in a recent episode. The show is fantastic overall btw. Strongly recommend subscribing open.spotify.com/show/4J8jzpL...

Michael I. Jordan rn

Smarter robots — inspired by dogs. Brown University researchers developed a new approach that helps robots determine which objects people want them to retrieve by interpreting both spoken language and human gestures.

I miss Brown and the CNTR. It’s the best place to work on exactly these kinds of questions!

That’s interesting. It might be revealing to also compare how the world models are distorted by media vs text. A-priori I’m not sure which distorts more: the average media diet or the average text diet? I would assume the former but who knows. The rate at which the distortion happens may also differ

This is a great idea for a research project.

The matrix is here :). Also the paperclip maximizers.