AI is often thought of as a black box -- there's no ready way to figure out what's going on inside. That's changing a little, in some eye-opening ways. In this essay I discuss how some researchers are gleaning what "beliefs" models are forming about their human interlocutors as they converse, and how those beliefs correlate to what the models say and how they say it. There are lots of policy implications!

https://www.theatlantic.com/technology/archive/2025/05/inside-the-ai-black-box/682853/?gift=GFVsesv7yr2B_e8FpoMTZ1YHTxzxzpBN6k87fsIFHxw&utm_source=copy-link&utm_medium=social&utm_campaign=share

I spoke with my old friend Jim Sciutto about the discovery that a copy of the Magna Carta at the Harvard Law School Library purchased for $26.50 in 1956 turns out to be a real one from AD 1300. (!)

If you’re just tuning in to the flurry of AI agent releases, here’s last summer’s short essay meant both as an explainer and a sketch of some real challenges for us all to take up. https://www.theatlantic.com/technology/archive/2024/07/ai-agents-safety-risks/678864/?gift=GFVsesv7yr2B_e8FpoMTZ3HZdGk7HxQRM9NGUsBu0p8&utm_source=copy-link&utm_medium=social&utm_campaign=share

@neilk fwiw https://www.theatlantic.com/technology/archive/2024/12/chatgpt-wont-say-my-name/681028/?gift=iWa_iB9lkw4UuiWbIbrWGc4qbnPc5fxHhdt-Xqt9ess&utm_source=copy-link&utm_medium=social&utm_campaign=share

Some ruminations on the weird phenomenon where my name crashes ChatGPT (!) https://www.theatlantic.com/technology/archive/2024/12/chatgpt-wont-say-my-name/681028/?gift=iWa_iB9lkw4UuiWbIbrWGc4qbnPc5fxHhdt-Xqt9ess&utm_source=copy-link&utm_medium=social&utm_campaign=share

On Sep 12 the Applied Social Media Lab (where I work) is hosting “Beyond Discourse Dumpster Fires: Strategies and Tools for Better Online Civil Space" to explore new ideas for healthier and more satisfying online communication.

I'm particularly excited to hear from @zephoria!

RSVP here, online via Zoom or (limited) in-person tickets in Cambridge MA: https://brk.mn/discourse

There will also be a recording available afterwards, so those who do not use Zoom can still watch!

I wrote a short piece about agent-based AI, its risks, and what we might do about them. It also serves as a preview of a new book, where I try to offer a picture of how humanity is both gaining power and losing control with the rise of the next wave of technology. More soon -- and thoughts welcome!

https://www.theatlantic.com/technology/archive/2024/07/ai-agents-safety-risks/678864/

The HLS Library Innovation Lab just released 360 years' worth of all US case law to the public -- a ten-year project with a ton of moving pieces. Can't quite believe it's done! https://vimeo.com/922493882

#Lifehacking is in pretty bad odor these days, and with good reason: a once-useful catch-all for describing how to make things easier has become a pit of #ProductivityPorn, grifter hustling, and anodyne advice wreathed in superlatives and transformed into SEO-compliant listicles.

--

If you'd like an essay-formatted version of this thread to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2024/01/25/today-in-tabs/#unfucked-rota

1/

We just had the panel prep call for this event on the Comstock Act and let me tell you, you will want to be there for it.

The Afterlife of Anthony Comstock: October 31, noon ET, on the internet. https://cyber.harvard.edu/events/afterlife-anthony-comstock

🚨 ACADEMIC JOB ALERT 🚨

Open-Rank TT Position in Media Ethics & Law | UMass Amherst Journalism Department

Posting is flexible regarding approach to the subject matter. Please spread widely within your networks. Feel free to direct questions to me, as I'm chairing the search. I'm happy to call UMass home. I ❤️ my colleagues and students. A lovely, exciting place to work.

https://careers.umass.edu/amherst/en-us/job/520380/open-rank-tenure-track-journalism-ethics-and-law

@academicchatter @communicationscholars @phdstudents #commodon #mediaStudies #academicChatter

Hello! The Berkman Klein Center for Internet & Society @ Harvard Univ. is hiring. We are looking for a capacity-building Managing Director to join a flourishing, intellectually robust community of faculty, students, fellows and affiliates from multiple disciplines and backgrounds, committed to studying and improving the digital space by building in it.

More info here --

https://cyber.harvard.edu/story/2023-07/hiring-managing-director

Some thoughts on the latest existential AI risk letter -- https://threadreaderapp.com/thread/1663623838943563776.html

@thomasrenkert thank you!

The World Wide Web turns 30 on Sunday. It's also dying, one link at a time.

My latest story for @medium: https://wordworking.medium.com/the-world-wide-web-is-turning-30-its-also-dying-one-link-at-a-time-f3494a47b339?sk=0a66d8955b5a9a675b5006b9d3342d75

ht @adactio
@Zittrain

@emilymbender really interesting. I think it’s possible that “emergent” is a term of art within Google - maybe that accounts for SP using it in a way that didn’t make sense colloquially? https://ai.googleblog.com/2022/11/characterizing-emergent-phenomena-in.html?m=1

My social network Minus got some grafs and a link in today's New York Times. Thanks to @bkc and @Zittrain for the mention.

https://www.nytimes.com/2023/04/19/technology/personaltech/tiktok-twitter-facebook-social.html

Come & work with me, Brent Mittelstadt and Chris Russell at the University of Oxford.

We are hiring a post doctoral researcher with technical expertise in fairness & interpretability in machine learning to work on trustworthy and accountable AI.

Deadline 24 April, 12:00pm UTC. Apply now! https://www.oii.ox.ac.uk/people/vacancies/

@dcrocker well yes, but that’s kind of a version of Poe’s Law

@jtlg I mean - with clever prompt engineering you could probably suss out something uncanny. But if the person judging the imitation game is not geared up to do that …?