Notices by Prof. Emily M. Bender(she/her) (emilymbender@dair-community.social), page 6

The AI Con book tour, next stops: I'll be in the UK next week, for the following two events:

July 15, 2025: in conversation with Catherine Pope at the University of Oxford’s Nuffield Department of Primary Care Health Services.

https://www.phc.ox.ac.uk/events/how-to-fight-ai-hype-and-create-the-future-we-want

July 16, 2025: in conversation with Guy Emerson at the Cambridge Centre for Data-Driven Discovery.

https://www.eventbrite.co.uk/e/the-ai-con-in-conversation-with-professor-emily-m-bender-tickets-1436256354419?aff=oddtdtcreator

Full details & links to recordings of some previous events:

https://thecon.ai/events/

"Men who sell machines that mimic people want us to become people who mimic machines. They want techno feudal subjects who will believe and do what they’re told. We, as people, are being strategically simplified. This is a fascist process."

https://organizingmythoughts.org/some-thoughts-on-techno-fascism-from-socialism-2025/

The AI Con book tour continues!

June 30: Alex in Berlin
July 1: Emily in Melbourne
July 17: Emily in Cambridge

Details: https://thecon.ai/events/

w/@alex

@cyberlyra Thanks.

I have lots of issues with the "AI skeptic" framing, starting with the fact that it sets things up as the question under discussion being "is AI real?" rather than "what harms are these companies doing?"

Sitting down for a lunch interview is definitely an interesting experience! Leaving the roughly two-hour conversation with George Hammond, I felt like we'd covered good ground but had no idea what of that would make it into the piece. In the end, I think this covers it pretty well, though I (of course) have a couple of quibbles/some context to add.

https://www.ft.com/content/9029cc1c-4a3f-42ca-9939-f3ef8e8336ae

>>

My remark that LLMs were "born shitty" was in the context of a discussion of Cory Doctorow's notion of "enshittification". That process involves something that starts of beneficial for consumers (or at least the consumers in focus). LLMs used as synthetic text extruding machines have no legitimate use cases and --- for all the reasons discussed in the stochastic parrots paper --- are prone to harmful outputs to boot.

>>

Got asked to review a book proposal for "A Guide to Prompt Engineering".

More accurate title: "A Guide to Poking at the Environmentally Disastrous Racist Pile of Linear Algebra Trained on Stolen Data and Exploitative Labor Practices to Produce Outputs You're Too Lazy to Learn to Evaluate"

To all communities currently being targeted as locations for data centers, the companies aiming to put those in deserve no benefit of the doubt. I'm really glad @timsteller.bsky.social is on top of this one.

https://tucson.com/news/local/subscriber/article_3ee5a61d-4d65-4ecb-8039-b05c1b02cda2.html

For everyone who was sad to miss it, the recording of the Data & Society event about THE AI CON (me and @alex) and EMPIRE OF AI (@karenhao) moderated by @tamigraph is now online!

https://datasociety.net/events/challenging-ai-hype-and-tech-industry-power

"Part of our task in the face of generative AI is to make an argument for the value of thinking – laboured, painful, frustrating thinking."

"[W]e also need to hold our institutions accountable. [...] university administrators are highly susceptible to the temptations of technology-driven downsizing, big tech donations, and the appearance of being on the cutting edge."

https://activehistory.ca/blog/2025/06/11/on-generative-ai-in-the-classroom-give-up-give-in-or-stand-up/

So the NYT then quotes, wtih an apparent straight face, the various excuses the model providers have for their systems getting it wrong:

This is honestly fully predictable from the tech bros at this point (though still baffling that they don't understand that papier-mâché of "good data" is still papier-mâché). What's more disappointing to me here is the failure of the journalist to call them on this bs.

There's so much to dunk on in this NYT piece and so little time, but I gotta start here: Some law profs at U Chicago did a study to see if the chatbots could answer questions based on specific materials and found, unsurprisingly, that they make shit up.

https://www.nytimes.com/2025/06/07/technology/chatgpt-openai-colleges.html

>>

Happy US book release day to me, my amazing co-author @alex and everyone else who can now open up their own copy!!

All the details on ordering, events, and news coverage at https://thecon.ai

https://dair-community.social/deck/@emilymbender/114365240078380868

@jplebreton @emma

Hi Emma & JP, I don't think I have anything in writing, but indeed "chain of thought" is still just repeated responses to "what's a likely next word", based on additional training data that looks like "chain of thought".

A couple of relevant MAIHT3k episodes:

https://www.buzzsprout.com/2126417/episodes/16095327-episode-44-openai-s-ridiculous-reasoning-october-28-2024

https://www.buzzsprout.com/2126417/episodes/15438320-episode-36-about-that-dangerous-capabilities-fanfiction-feat-ali-alkhatib-june-24-2024

"We cannot afford to replace the critical tasks of federal workers with models that completely make stuff up. There is no substitute for the expertise of federal workers handling sensitive information and working on life-critical sectors ranging from health care to immigration."

https://www.scientificamerican.com/article/replacing-federal-workers-with-chatbots-would-be-a-dystopian-nightmare/

I've now seen a second instance of a quote attributed to me in a news site I've never heard of (and certainly was not interviewed for). In general, if the quote includes the phrase 'AI' with no scare quotes or other distance, that's a good sign it's not really from me.

I've written to ask for a retraction, so we'll see. I suspect this is yet another case of someone posting the text extruded from an LLM as if it were news. It's appalling.

What to do about a divergence between "public" and "expert" opinion about "AI"? Some might think that we need to educate the public so that they come to understand more how experts do. I disagree: expertise in building these systems is far from the only relevant expertise.

>>

Tomorrow!

~~

On the next Mystery AI Hype Theater 3000, @mmitchell_ai (Our first repeat guest! Welcome back Meg!) joins me and @alex to talk about “AI” “agents” -- what that means, what the salesmen want you to think it means, and what to actually be concerned about.

Catch the livestream here:

Monday March 31, noon Pacific
https://twitch.tv/dair_institute

Absolute highlight of this trip: **finally** meeting @alex in person. Since 2020, we have:

1. Co-authored multiple peer-reviewed papers
2. Co-authored op-eds
3. Co-hosted a podcast now at 53 episodes
4. Written A WHOLE BOOK (thecon.ai)

And now we've finally met :)