Ro Encarnación

@penn-mediated.bsky.social is at the #CJ2026 and would love to chat with folks. Come to our workshop where you can run your own LLM audit (today at 3:30) or I'm giving a talk tomorrow on the broader needs for scientific and longitudinal monitoring of LLM politics (in Parallel Talks 1 at 1:30).

Looking forward to presenting "𝗪𝗵𝗲𝗻 𝗟𝗟𝗠𝘀 𝗚𝗲𝘁 𝗔𝗯𝗼𝗿𝘁𝗶𝗼𝗻 𝗟𝗲𝗴𝗮𝗹𝗶𝘁𝘆 𝗪𝗿𝗼𝗻𝗴: 𝗔𝗻 𝗔𝘂𝗱𝗶𝘁 𝗔𝗰𝗿𝗼𝘀𝘀 𝗨.𝗦. 𝗦𝘁𝗮𝘁𝗲𝘀" tommorow at the Trust and Safety Research Conference! Catch this talk at the tail end of the 10am session in the Fisher Room. I'll share a post-audit update :) #TSRConf Check out the paper here.

Look, ma, I'm in a magazine!! 🤩✨ I got to speak with @motherjones.com about #PublicAI and how access, accountability, & sustainability are about more than just shareholding. www.motherjones.com/politics/202...

exactamente!! La ciencia lenta respeta el proceso y el tiempo que toma investigar problemas complicados.

Can't wait to read this!

It’s unfortunate that gen AI providers can continue to dismiss evidence of something going awry in their models bc the “newest” version wasn’t used (even if the model used is still in prod). Research can’t keep up and newer models cost more per token/API call. This era of AI evals is expensive!

Please reach out/message me if you’d like to join!

Curiosity mixed with rage is really undersold as a motivator for sociotechnical research.

Everyone who says that you need to learn and use AI heavily right now to avoid being permanently left behind is telling you that they believe the future will feature profound and inescapable inequality and they’re okay with that.

Thank you so much for sharing our work! 🙌

So happy for my home district 13!!

Policymakers by and large don't read academic papers! If you're looking for policy impact, you need to show up. First you need to learn _how_ to show up. Excited to cohost this #FAccT2026 craft session with @rone.bsky.social, @narijohnson.bsky.social, Cynthia Bailey, and Danaé Metaxa.

Yet another reason why a 97.8888% AI eval score is meaningless. Even just ONE error like this is insane.

At this point, I would rather see a flyer made with clip art and typed with Papyrus or Comic Sans than have to look at another AI-generated flyer.

The move from open regulatory processes that engage the public to clandestine, executive decisions about technology isn’t deregulation. It’s just a rose by another name. Folks, the brilliant @alondra.bsky.social

Here we go with dragnet practices again, courtesy of data broker loopholes (and data brokers in general).

Thanks for continuing to read, cite, and share this paper. It remains controversial in some corners of the discipline, but I stand by it completely. Many Black folks don’t use terms “liberal” and “conservative,” which makes the liberal-conservative scale a noisy tool for studying Black politics.

AI companies are paying screenwriters, lawyers, and other white-collar professionals to produce the training data needed to automate their jobs. I spoke with more than 30 workers about conditions inside this fast-growing and extremely secretive new gig economy.

It's been a journey of nearly 3 years, but I'm very excited to announce the CNTR AISLE Portal! 🚀 cntr-aisle.org It’s a new way to review and evaluate the 1,000+ AI bills introduced in the U.S. over the last three years. Check out the Bill Library and our Profiles#AIPolicy #OpenData

In Portland, the Water Bureau is piloting an algorithmic pricing system to automate bill discount levels. I've reported extensively on this otherwise ignored project. Lawmakers in CA have introduced several bills to regulate algo pricing, a very nuanced area. My reporting and more on CA follows:

With such low AI literacy rates (and low literacy rates in general -- another convo on critical thinking) and inequitable information access, can we move beyond putting the onus on consumers of these tools to be more responsible?

Meanwhile, ChatGPT guided this teen on how to maximize his drug “trips” until he died from it. www.sfgate.com/tech/article...

We're getting closer and closer to making tech dystopia novel, "The Marriage Act" by John Marrs, a reality: www.washingtonpost.com/business/202...

this is irresponsible reporting. a neural network can't comment on why it produced CSAM. a neural network can't be held answerable for generating CSAM. accountability lies within the fascist ceo and the team willing to work for him www.reuters.com/legal/litiga...

Can public involvement in AI evaluation improve the science? Or does it compromise quality, speed, cost? In @pnas.org, Megan Price & I summarize challenges of AI evaluation, review strengths/weaknesses, & suggest how participatory methods can improve the science of AI www.pnas.org/doi/10.1073/...

I'm convinced that many current "safety guardrails" in generative AI systems are merely financial and political risk mitigation strategies. They are meant to protect companies first, not users.

When it comes to AI, ignore what Sam Altman says and instead listen to @abeba.bsky.social @timnitgebru.bsky.social @olivia.science @irisvanrooij.bsky.social @karenhao.bsky.social

Oops, $1.5 billion to Prime subscribers! Another $1 billion to the FTC.

Amazon has to pay $1.8 billion due to deceptive Prime subscription practices and change those same practices as part of an FTC settlement. It’s time to screenshot for posterity and wait for Amazon to use the settlement to raise Prime. Penalizing customers for what they did TO them.

The dark patterns in abortion crisis pregnancy center websites are a special level of insidious and predatory. The "For Men" section in some of these websites goes to show how far they'll go. This screenshot is an example. The "Your" here is not referring to the pregnant person.

Heavily considering replacing bedtime stories with dramatic readings of research papers. Work smarter not harder yk?

Google is also trying to lock in schools to a particular set of ideas that it claims to be based on "learning science" with its new LearnLM/Guided learning bot. It's a *theory lock-in* to a very attenuated version of learning theory. blog.google/outreach-ini...

Are you an HCI researcher from or who studies the Global Majority? Reviewed research about Global Majority for HCI venues? @farhana-shahid.bsky.social & I are conducting research on peer review experience of research by and about Global Majority. participation form: docs.google.com/forms/d/e/1F...

How can we ensure that #AI is ACTUALLY helpful for the BIG challenges we face, like #ClimateChange and #WorldHunger? We need to listen to the experts who understand these problems best! That's why @aspendigital.bsky.social is surveying people in #FoodSecurity: www.aspendigital.org/event/food-s...

I'm incredibly fortunate to have had the opportunity to work with this team. Truly one of the best collaborative experiences I have had to date (special s/o to our MVP @mkgerchick.bsky.social for leading this)! Check out Marissa's talk on our paper "auditing the audits" if you're at #FAccT2025! ⬇️

The leeway permitted by "automated decision systems" alone should terrify anyone regardless of their views about "AI".

I know the FTC is already fighting this but I’m gathering more receipts for my own purposes. Amazon did try to subscribe me again after I cancelled. I then deleted all my payment options. So why did they try charging my card this morning?? Looks like they don’t delete your payment info either.