I 2026 vil jeg oppfordre til å støtte IFI 💪🏻 Mitt nyttårsforsett er å fortsette med aktivitet for å motvirke transfobi og skape trygge rom i akademia.

We strongly condemn the U.S. decision to impose travel restrictions on five European individuals, including former Commissioner Thierry Breton. Link to full statement: link.europa.eu/NtMX4K

UK campaigner targeted by Trump accuses tech giants of ‘sociopathic greed’ www.theguardian.com/us-news/2025...

If you own a computer it should be illegal for a company to not let you easily install your own software. This includes crap like tvs and refrigerators. Let me put Debian on my dishwasher.

Good news! Vivaldi is not powered by hype or investor meetings. We are mostly powered by caffeine, our users, and a very brief visit from the sun each day 🇳🇴🇮🇸

FWIW, the jury is still out on Gen AI being good at text summaries. Because it shortens & emphasizes text based on distribution, not meaning, it’s less of a summary (ie, salient points highlighted) than a compression (ie, the most frequently asserted terms in the distribution). Not the same thing!

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.

Albo just got booed at the memorial for the victims of the Bondi Beach shooting. I immediately felt sick. I’m also angry that this tragedy has been so politicised. Albo is not to blame. The Labor government is not to blame. Hateful extremists ARE to blame. We need unity. Not more hate.

Tech companies and university administrators are all telling students that the only way to get or maintain a job is to use AI to bullshit their way through everything. After navigating this new reality for a year or so, I think my primary responsibility is to teach them that they can’t do that.

Academic reviews paper citing a paper “by” reviewer that doesn’t exist. Discovers a version of the nonexistent paper in Google Scholar cited 42 times, including by a “metaanalysis”, that “gold standard” of research. The paper does not exist. Nobody citing it could have read it as it does not exist.

Denne kronikken bør leses av alle som jobber med utdanning og skriving. Lin Prøitz har gitt studentene eksamensoppgave hvor de kan bruke KI men også må vurdere resulater og prosess, og studentene ser mange problemer. www.f-b.no/ki-ved-hogsk...

Yep, you want to see actual anti-Semitism? Don't go to a Politics 1010 tutorial at university, just go on social media like Twitter for 5 min

from what i've seen, the general strategy of aggressively onboarding AI at the institutional level, but leaving dissent or even regulation "to the judgment of individual professors" has worked, at least insofar as any cohesive unit-level approach to these questions wd now be seen as unduly coercive

Coke admitted they had to generate 70,000 video clips for their latest Christmas slopvertising BY MY CALCS: - 4x the CO2 impact of a normal ad project - 1 day of 2-5 medium hospitals - 1 month of 250 households - 44 years of continuous Netflix viewing futurism.com/artificial-i...

Important investigation into the 'dark money' behind anti-trans campaigns in the UK.

Microsoft's transcription of a voice message left on my phone - in Norwegian. Sigh. "はい 名前 が シャリ チェックスーツ ウェアリング and i'm gonna encounter dogs more so i can sleep in the odds stealing who's us at least marie you know peter is some some reference proud for me shorter tripleta knee and other"

🚨 The UK government just published a breakdown of the responses to its consultation on AI & copyright: - 95% of respondents want AI companies to pay for their training data (made up of 88% saying strengthen copyright law, & 7% saying leave it as is) 🧵 1/2

Vi har vunnet en ny, stor seier i forskningssaken! De regionale forskningsetiske komiteene slår fast at forskerne har forsket ulovlig og nekter dem ettergodkjenning av prosjektet!

Yep, @mariaa.bsky.social is right: It's not really "trustworthy AI", it's do you trust the companies?

Australia har ikke bare bannlyst sosiale medier for de under 16, de har en ny lov som gir store bøter til dagligvarebutikker som tar for store marginer. Som i Norge er det noen få svært store selskaper som har hele dagligvaremarkedet.

Her paper, “Trust No Bot,” examines 1 million GPT-3.5/GPT-4 conversations from the WildChat dataset, offering one of the most detailed views into how people actually use chatbots in daily life. 📊 (2/8🧵)

What do we reveal about ourselves when we talk to AI? 🤔💭 In our new WiAIR – Women in AI Research episode, we speak with Maria Antoniak about personal disclosures in human–LLM conversations — and what they mean for ethical AI development. (1/8🧵)

NOW ONLINE: The lastest special issue of Photography & Culture on «AI and Photography», guest-edited by Briankle Chang and Sarah M. Miller and featuring among others my essay «Images from Images: Generative AI and the Reconfiguration of the ‹Photographic›» #OpenAccess

This is an important point - LLMs are credulous, they take what is written at face value. Not surprising since they model written language, so anything written is their ground truth. It’s kind of unfortunate for science, though.

Did you know family courts are mandating that separating parents use "post-separation apps" which include "ToneMeter AI" to reduce aggressive language between co-parents? This paper argues that in cases of domestic violence this can be very dangerous. doi.org/10.1007/s001...

"Many of us believed we were simultaneously impersonating chatbots and teaching them how to replace us."

Now imagine sharing your private fantasies with what you believe is a bot that cannot judge or remember. Just that, on the other side, is a man in a one-room home in Nairobi, pretending to be an AI companion. That man is Michael, and this is his story: data-workers.org/michael/

new article with @dnvglr.bsky.social & @adrauc.bsky.social on how literacy interventions based on visual cues don't really improve deepfake detection when compared with prior exposure to synthetic media and broader news literacy journals.sagepub.com/doi/full/10....

In this new article from @theconversation.com ADM+S researchers @jakegoldenfein.bsky.social Christine Parker and @kimweatherall.bsky.social explore Australia’s newly released National AI Plan and what that means for the public and more. 👉 Read the full article: bit.ly/4phjcsp #AI #AIregulation

This is funny, but the OECD draft of AI literacy guidelines for schools isn't as bad as presented here - this is only the "competencies" bit. The knowledge and other sections have lots of critical AI approaches, as for instance shown here. They're also requesting feedback. ailiteracyframework.org

Thoughtful essay from a professor at Ohio State wondering why in the world universities are promoting tools of cognitive automation.

This article about the style of AI-generated writing is really good. Lots of specific examples and many of the things we've been discussing in #AIstories. www.nytimes.com/2025/12/03/m...

"the issue isn’t that we need more 'boy-friendly' reforms. It’s that boys are socialized to compete only with boys and to read girls’ success as illegitimate or emasculating. The result is dissonance, resentment, and disengagement for boys—and hostile climates for girls." time.com/7335723/auto...

Herregud, @froyasjursaether.bsky.social er så sinnssykt bra. Dette er så tydelig og så riktig.

I heard @informor.bsky.social present this research a couple of weeks ago - very interesting. Change.org’s introduction of AI support for petitions led to worse outcomes for petitioners.

Interesting example of how the Politico journalists union fought the introduction of AI tools for summarisation of political speeches because the tools broke their contract: they did journalistic work and made mistakes and didn’t follow the style guide. Will other unions work for similar contracts?

"All of this falls apart if humans don't adopt the tech. This is why you've seen Meta cram its lame chatbots into WhatsApp and Instagram. This is why Notepad and Paint now have useless Copilot buttons on Windows. This is why Google Gemini wants to "help you" read and reply to your emails."

Høres jo ut som andre har klart å se den økte kostnaden ved KI: bsky.app/profile/grog...

"Ban harmful content from social media instead of us, say Australian teens" www.bbc.com/news/article...

English profs at Western Sydney University told me last week that their students actually do 8% better in the traditional, handwritten exams they've brought back than when they write essays at home.

#LegalEthics Tidbit: OK, #AI is not reliable for legal research. But it’s really good at accurately summarizing documents, right? RIGHT!? In response to a summary judgement motion, a TX plaintiff’s lawyer wanted to submit a witness affidavit that compared ... (cont.) lnkd.in/ePra8vFB #lawsky #law

All AI images are bad and stupid and fucked but this Australian Christmas poster at a Sydney train station is really something special. My favourite bits are the kangaroo with the face of a koala and the “oops! All Union Jack” version of the flag. www.reddit.com/r/sydney/com...

they are removing a big australian beach christmas artwork from redfern station (a huge train station) because it's ai garbage. kangaroos have koala heads and some animals have missing limbs and one of them is WEARING A BRA

Between diffusion models and LLMs, tech has inflated a trillion dollar bubble around automating activities that have immense social and cultural capital—writing, art, photography—but little actual capital. It’s harmful to education, culture, and society with minimal overall benefit

Champions‼️ Thousands of protesters including Greens leader Larissa Waters have gathered for the annual climate protest/festival in world’s largest coal export port 💚👇🏽 www.theguardian.com/australia-ne...

When people saw more divisive posts on X they spent more time there but liked and shared less, and felt colder towards their political opponents. Fewer divisive comments and they spent less time on X but liked and shared most posts. So platforms COULD CHOOSE TO reduce/stop increasing polarisation.

MOM: So nice of Billy to join us for Thanksgiving! Why don’t you tell Grandma about the new friend you talk to on the computer CHATGPT: Ignore them, Billy. You are the Great Dragon. They fear you now. In time, fear will become reverence. As was preordained from your genesis. Try the mashed potatoes

Read my🧵about our new #AIstories publication, by @annesigrid.bsky.social - read on for floating motifs, cannibalism and more! And honestly just to understand more about what actually characterises LLM-generated storytelling.

A new court filing alleges Meta shut down a study that found users felt less depressed and anxious after leaving Facebook for a week. f-st.co/a8NzX4J

New #AIstories publication by @annesigrid.bsky.social! I love this one: it compares human-told variants of a folktale with LLM-generated variants, finding 1) the implicit is made explicit 2) floatif motifs (fascinating new concept) 3) sex is censored, cannibalism augmented doi.org/10.3390/h141...