What happens when #AI shapes human disagreements? Join us on Tuesday, April 21 for a seminar with @jonathanstray.bsky.social of the Center for Human-Compatible AI at @ucberkeleyofficial.bsky.social to explore research offering a glimpse into the turbulent future of AI-mediated conflict. RSVP ⤵️

I agree there's a real argument AI will lead to homogenization. It's also striking that, until very recently, everybody was wringing their hands about "fragmentation" and "filter bubbles." Without a theory on how much diversity vs. commonality we want in media, we will remain confused and reactive.

Writing about attested.network, an open spec for decentralized proof of payments on ATProtocol. It builds on what we learned making atprotofans.com. Draft is up and feedback is welcome.

Stray's analysis reveals the core measurement problem: external researchers can only count hateful posts produced, whilst X claims to reduce hateful impressions through algorithmic suppression. Without internal data, we're arguing over different metrics entirely.

Excited to see my #AtmosphereConf talk up on a legacy platform:

Did "hate" go up on X/Twitter after they changed their content moderation rules post-Musk? What is "hate" anyway, and what data do we actually have? I've got all the charts and analysis I know you crave! New by me for @techpolicypress.bsky.social www.techpolicy.press/what-are-the...

As far as I can tell, the new X algorithm just completely ignores who you follow when selecting out-of-network posts in the For You feed. This might be why it feels, to some people at least, less personal or cozy. www.greenearth.social/p/making-soc...

But let's be honest: there's a large cadre that basically cheers on chasing off any lib/centrist/academic who's the punchbag of the day. There's a culture of saying "fuck off back to X, then". And the anti- bedtime leftists set too much of the culture.

Can feed algorithms build community? We analyzed the old and new Twitter source code to figure out why it seems like a less friendly place than it used to be. We're designing GreenEarth to connect people instead. www.greenearth.social/p/making-soc...

Game theory does not have a way to distinguish "conflict" from "competition." I think the difference is whether people are using destructive moves (e.g. murder) to win the game. Such moves are defections in the meta-game of peace and security.

It now seems likely that machines will soon become much smarter than humans. But will our superintelligent machines finally bring an end to war? Here's where current AI fails, and what we can do about it. www.betterconflictbulletin.org/p/conflict-s...

Folks have been reaching out about how to support Graze beyond our core product. We think that support needs to go beyond Graze. That's why we've teamed up with our direct competitors @skyfeed.app and @blueskyfeedcreator.com to jointly support all our work:

Hey @graze.social @skylight.social @pfrazee.com wanted to make sure you'd seen this.

Hey #atmosphereconf we just proved that the right feed algorithms can reduce political polarization. And now we're building open-source recommender infrastructure that you can use for your ATProto app. rankingchallenge.substack.com/p/its-possib...

Could social media make us less polarized instead of more? We tested 5 algorithms on 3 platforms with 10,000 people for 6 months during the 2024 election, and found that the answer is yes. 🧵

I submitted the Big Paper and now what do I even do with my life

Honestly this weekend was when OpenAI jumped the shark for me. I was willing to extend them benefit of the doubt right up to the point where they signed a contract with the government that allows autonomous weapons, while telling everyone that it prohibits them.

As far as I can tell, there is nothing in OpenAI’s new contract that prevents the Department of War from building fully autonomous killer robots with their tech. www.betterconflictbulletin.org/p/openai-jus...

At the International Association for Safe and Ethical AI (IASEAI) conference in Paris, listening to French electronica. Seems appropriate. www.youtube.com/watch?v=N65s...

Since purchasing X, Elon Musk has shaped the platform to satisfy his own interests. But what can we observe about how his interventions have favored one set of politics over another? Asking this question may help us identify where bias may emerge across all platforms, writes Jonathan Stray.

This is why I work on AI & political conflict. Or one reason why.

Update: I met Joanna. She's a real person! Who really did comment on every single paper! And had trouble downloading the PDFs. I stand corrected. Sorry Joanna!

Has Twitter/X gone right wing? Let's answer this with data! First post of a three part series on auditing the politics of a platform from the outside. In part 1, I look at shifts in X's user base. www.betterconflictbulletin.org/p/what-are-t...

I kinda like that someone wrote a bot that commented semi-sensibly on EVERY SINGLE paper at #IASEAI.

The digital services act - an EU transparency law - might actually be working to give researchers access to the data they need to understand platform effects.

Look, they're not called Tweets anymore. And they're not even called posts. From now on, they're SpaceXeets. Violators will be demonetized.

The arc of history bends towards justice, but only because millions of people are working the problem.

Open Social Web Series 🌐 Building on the AT Protocol means we don't have to build alone 🙌🏽 Mentions: @jonathanstray.bsky.social @graze.social @stream.place @gandersocial.bsky.social @transrights.northsky.social @skylight.social #OpenSocialWeb #SkylightSocial #BuildWithTori

I just rejected a paper about AI slop because it was AI slop. I suppose one day AI will write good papers, but today is not that day.

Either ICE has started cruelly arresting beloved community members, or ICE is merely enforcing existing laws to remove murderers and rapists. Both of these are only half-right. Whatever side you’re on, you probably have something to learn. www.betterconflictbulletin.org/p/annoying-f...

Media is fiction when everything is recorded first, selected later. Media literacy is understanding that the popular is extreme. Trustworthy sources are better at modeling and communicating the true distribution of events. Narratives define what counts as an event.

I'm reverse engineering the Twitter algorithm from the source code and experiments. On my own For You feed, only 16% (1 in 6) of posts are from people I follow. This used to be set at 50% in old Twitter. So that's one reason it feels like we've lost our communities there.

I RTed this this morning before reading it (my partner hates that I bookmark that way!), because I hadn’t seen anyone else talking about solutions/ what’s next. I’ve read it now and think solution 2 is both a good and a plausible next step — states choose & perform their own migration strategy. 1/

What's the endgame for ICE protests? As far as I can tell, I'm the only one who has publicly proposed the use of targeted civil resistance to force negotiation of an immigration policy compromise that might appeal to all Americans. Here's my analysis... www.betterconflictbulletin.org/p/what-would...

I am thinking a lot today, like many days but more so, about what information an algorithm should show you during a crisis. What will inform, calm, show us all a way through regardless of where each of us stands?

Math Sky, I just learned about Robinson's infinitesimals and I am shook, just shook to learn that you can do calculus entirely without taking limits. Newton and Leibniz were right, they just couldn't prove it.

Ah crap they asked me to area chair.

Become ungovernable

Optimizing for engagement alone can lead to bad results for social media and LLMs. So what's the alternative? You can just ask your users what you actually want to know! Here's Pinterest explaining how they did just that... medium.com/pinterest-en...

Anyone got a take on this study suggesting COVID vaccines (both mRNA and other) increase the risk for some types on cancer? It seems well designed and plausible, but it does have an editor's note saying "concerns have been raised." link.springer.com/article/10.1...

The Sagrada Familia is every bit as awe inspiring as you’ve heard.

I’ve just discovered King Tubby and my understanding of music history is changed forever. He invented dub and pioneered the producer-as-creator… in the 1970s

Ok. I love this conference but my first and probably biggest criticism of #buildpeace2025 is that, although “plurality” is a recurring theme, pretty much everyone here has strongly progressive politics. A lot of talking *about* the other, not *with* them.

Ola! I’m here in Barcelona for #buildpeace26 Who else is here? I see @loreleikelly.bsky.social

Reviewing papers for IASEAI 26 is exciting! Lots of groundbreaking AI safety work! But I can’t help shake a certain imposter syndrome… I am now literally a gatekeeper in my field.

I saw Bugonia and wow. Not for the faint of heart, but incredible.

So I met this guy who’s meditated 4000 conversations between Israelis and Palestinians since October 7th. And he’s building an AI powered platform for difficult conversations. Here’s what he’s learned. www.betterconflictbulletin.org/p/what-you-c...

This date could have been a sext.

ATProto is just a 21st century NNTP, amirite?

@lawfaremedia.org where are you on this? I need expert Sandwich Guy legal analysis. The fate of our nation depends upon it.