Responsible researchers have the privilege to unhitch themselves from the "AI" brand at any time. Are you studying wildfire smoke or deforestation using satellite imagery? That used to be "remote sensing" and you can call it that again. Signal processing, image segmentation, structure-from-motion.

This. In the past several years, many have branded their work "AI" for the financial benefits, despite the intense toxicity of the people and tacit assumptions at the center. As communities become conversant in critique and refusal, fragility is on full display as researchers cry out "not all AI". 🧵

That last point, about learning in community, really resonates with me. It connects with what I say in this piece for UNESCO: www.youtube.com/live/l-OWi6V... (Written version available here, pp.41-45: doi.org/10.54675/KEC... ) >>

These people have been told that the chatbox is an everything machine. The way out can't be to provide a better everything machine because that can't exist.

Every act of refusal can help pave the way for others to refuse as well.

Really enjoyed this fantastic piece by @sonjadrimmer.bsky.social and @cnygren.bsky.social Here are few favorite pull quotes of mine + one little quibble at the end:

Professional societies keep beclowning themselves buying into a lie about what an LLM "summary" is. They are inherently counterfeit: not an epistemic product of the ideas in the source, but summary-shaped text linguistically based on *other* works (in the training corpus) that use related language.

Every act of refusal can help pave the way for others to refuse as well.

Really enjoyed this fantastic piece by @sonjadrimmer.bsky.social and @cnygren.bsky.social Here are few favorite pull quotes of mine + one little quibble at the end:

same. and if you're like "but what about [a predictive model that is not an LLM and is actually useful]" then maybe talk about it as a predictive model that does something useful and don't call it "ai" If it's a different thing use different words to describe it

This quote from @olivia.science @marentierra.bsky.social and @irisvanrooij.bsky.social is in my recent free one-page zine v3 called "What Does 'AI' Actually Mean? Definition and Synonyms" (and Emily Bender is also cited in v2 and v3!) Zine: padlet.com/laurenUU/antiAI Paper: doi.org/10.5281/zeno...

I'm really glad to see mainstream news attention to the problem of chatbots using "I/me" pronouns (there is no "I" there), but also frustrated to see Askell's flat out lie just platformed here. www.nytimes.com/2025/12/19/t... by @kashhill.bsky.social >>

Think of AI instead of cigarettes here. Nothing is inevitable.

I'm really glad to see mainstream news attention to the problem of chatbots using "I/me" pronouns (there is no "I" there), but also frustrated to see Askell's flat out lie just platformed here. www.nytimes.com/2025/12/19/t... by @kashhill.bsky.social >>

So I think the move for the domain experts in this situation is to resist the urge to hedge along the lines of "Well, 'AI' might have its uses elsewhere, but here it's bad." >>

We do not have to offer and I believe indeed should not offer any benefit of the doubt: Anything sold as "AI" should be presumed ineffective, exploitative, environmentally ruinous, bigotry-amplifying and otherwise until proven otherwise. /fin

Reading "Stop Treating ‘AGI’ as the North-star Goal of AI Research" by Blili-Hamelin et al and particularly struck by this passage: icml.cc/virtual/2025... Short 🧵>>

Reading "Stop Treating ‘AGI’ as the North-star Goal of AI Research" by Blili-Hamelin et al and particularly struck by this passage: icml.cc/virtual/2025... Short 🧵>>

These people have been told that the chatbox is an everything machine. The way out can't be to provide a better everything machine because that can't exist.

Altman and Ive have ruined San Francisco. @emilymbender.bsky.social and I take on the ridiculous marketing video which came out of OpenAI several weeks ago in Ep. 58 of our podcast.

Anthropic C-suite folks are ridiculous, but unfortunately also empowers to impose things on others based on their batshit views. www.404media.co/anthropic-ex... Short 🧵>>

NEW: Mystery AI Hype Theater 3000, Episode 68: This is What Algo-cracy Looks Like Tech bros say automation can strengthen democracy. @alexhanna.bsky.social and @emilymbender.bsky.social unpack their flimsy arguments about the democratic process. www.buzzsprout.com/2126417/epis...

This one is great for showing how much fun we have with the livestream audience. We'll be back with more livestreams in January!

🚨 THE FINAL FOUR 🚨 We’re nearly at the end! 32 candidates have been narrowed down to a nefarious final four. But who will move on to the finals? You must decide! 🗳️ Cast your ballot: twsu.forms.app/wpit25-semif...

This is 100% infuriating.

I don't know how many times this needs to be said, but apparently a lot: If you are not in a position to verify the accuracy of some synthetic text, the synthetic text is not useful/has a high potential to be misleading. This is one of those cases.

And, for fun, here’s a bonus audiogram!

Mystery AI Hype Theater 3000, Episode 68: This is What Algo-cracy Looks Like www.buzzsprout.com/2126417/epis... @alexhanna.bsky.social and I explore what democracy looks like in the funhouse mirror of tech bros’ imaginations. Thanks to Ozzy Llinas Goodman for production!

Accessibility without verifiability isn’t clarity—it’s risk transfer. AI summaries don’t simplify truth; they abstract accountability. If readers can’t validate claims, the output degrades from aid to liability.

I don't know how many times this needs to be said, but apparently a lot: If you are not in a position to verify the accuracy of some synthetic text, the synthetic text is not useful/has a high potential to be misleading. This is one of those cases.

This one is great for showing how much fun we have with the livestream audience. We'll be back with more livestreams in January!

Also available as video on PeerTube: peertube.dair-institute.org/w/1SXevXYdaD...

Mystery AI Hype Theater 3000, Episode 68: This is What Algo-cracy Looks Like www.buzzsprout.com/2126417/epis... @alexhanna.bsky.social and I explore what democracy looks like in the funhouse mirror of tech bros’ imaginations. Thanks to Ozzy Llinas Goodman for production!

Shame on the ACM. They should know better. In fact, here's a fun exercise to the reader: How many points from the ACM's own code of ethics does this move violate? www.acm.org/binaries/con...

@cfiesler.bsky.social covers all of the important points in her thread (read the whole thing) but the fact that they are covering up the abstract (i.e. what the authors themselves wrote to give readers a sense of what's in the paper) with slop is telling, I think. >>

This is 100% infuriating.

🚨 THE FINAL EIGHT 🚨 We’ve made it to the quarter finals — only two days away from crowning a winner for Worst Person in Tech 2025. Don’t miss your chance to vote! 🗳️ Cast your ballot: twsu.forms.app/wpit2025-qf

The University of Washington's Center for an Informed Public is looking to bring on new postdoctoral scholars to join our research community starting in 2026-2027. Come collaborate w/ our amazing, interdisciplinary team working to understand & help people navigate our complex information ecosystem!

Was really challenging to participate in this, but I think I was able to pen something real, important, and personal to my experience in AI. Remember that GenAI is not *all* of AI nor all of what it should be. www.technologyreview.com/2025/12/15/1...

"We envision a resistance that is...a repudiation of the efficiencies that automated algorithmic education falsely promises: a resistance comprising the collective force of small acts of friction." "How to Resist AI in Education" by me & @cnygren.bsky.social www.publicbooks.org/four-frictio...

Since then, she and Alex Hanna have continued their excellent podcast Mystery AI Hype Theater 3000, and published a book (The AI Con) that discusses many of the threads we touch upon in this episode! The book: thecon.ai The podcast: www.dair-institute.org/maiht3k/

Back in the summer of 2023, we were honored to be joined by @emilymbender.bsky.social to discuss the rhetoric of AI hype and her critical work contextualizing Large Language Models for the public. www.reverbcast.com/podcasts/202...

🚨 ROUND TWO STARTS NOW 🚨 We’re down to the final 16 contestants as you decide who will be this year’s Worst Person In Tech! Each day of this week new matchups will drop until we choose the winner on Friday. 🗳️ Cast your ballot: twsu.forms.app/wpit25-round...

🎧TODAY: tune in to the last Mystery AI Hype Theater 3000 live stream of 2025 🎧 Time Magazine is looking back and a year of "AI" and so @emilymbender.bsky.social and @alexhanna.bsky.social will be doing the same! TODAY, Dec 15, 8pm BST/3pm EDT/12pm PDT twitch.tv/dair_institute

They all suck -- we might as well have a little fun in considering just how badly they suck.

This post (+ the one after it, but starting with this one because context is important) has gotten surprising amounts of attention from folks who think that students are necessarily lazy and malicious. >>

This was interesting one to be interviewed for -- and I'm looking forward to seeing it at Sundance!

Today!

From information superhighway to fake "information" and no highways.