Excellent piece on what it means when we let agents take over: "the beginning of a world in which people are less in the loop and more on call, less empowered by the technology and more enslaved by its tempo" Highly recommended read! www.noemamag.com/ai-agents-ar...

"Ich erhielt ein Angebot, meine Forschungen in einem militärischen Kontext fortzusetzen. Das kam aber aus ethischen Erwägungen für mich nicht infrage" - na dann lieber Geld von Epstein nehmen... www.zeit.de/wissen/2026-...

Unlikely Allies: 20min geht eine Kooperation ein mit The Economist und serviert seiner Audience 2 Mal wöchentlich Economist-Artikel. Spannend, dachte ich. Bis ich merkte: Maximaler KI-Slop! Maschinell übersetzte seelenlose Pampe. Unlesbar. Links im nächsten Post ->

An LLM is trained on "right answers", i.e. the next word in the text that's been censored. In real use, that target vanishes: there's no correct next token when a person says, “hello” to the model. @benjaminjriley.bsky.social & @colin-fraser.net buildcognitiveresonance.substack.com/p/how-to-out...

Eine junge Frau hat mir erzählt, dass sie am Wochenende an einer KI-Konferenz war, wo ein Speaker erzählte „KI wird uns alle umbringen“. Seither kann sie nicht mehr schlafen. WTF stimmt nicht mit solchen Typen?

So, ChatGTP Health answers the same heart longevity-grade question with a score swinging between an F and a B. Claude made it vary between a C and a B-. Anthropic said chatbots have "inherent variation in outputs". yes, inherent, like throwing a dice. www.washingtonpost.com/technology/2...

The "misconception in the AI world that reasoning is a separate skill that's easy for humans" fosters a "widespread belief that AI will soon be superhuman at things like scientific research or operating a company or even a government". Great video by Arvind Narayanan youtu.be/-Ty4BXrASFU?...

KI-Plattformen brauchen unsere Kreativität. Doch je erfolgreicher generative KI wird, desto mehr zerstört sie ihre eigene Grundlage. Es geht um Model Collapse, externalisierte Kosten und die Frage nach Verantwortung. Ein paar Gedanken dazu hier: dorotheabaur.ch/texte-und-me...

“The idea is to put ChatGPT front and center inside software that scientists use to write up their work in much the same way that chatbots are now embedded into popular programming editors. It’s vibe coding, but for science.”

Haha, Accenture - hat jahrelang Millionen mit SEO- und Content-Marketing-Beratung. Nun nutzt der CEO die Plattform nicht mehr, von der diese Strategien abhängen. Zitat: "Überall dort, wo ich früher gegoogelt habe, nutze ich heute KI" 🤦‍♀️ www.20min.ch/story/untern...

Wunderbar. War kürzlich auf einer Party in Zürich, wo sich alle moralisch erhaben und grün gaben. Über die Festtage flogen sie wahlweise nach Tansania, auf die Philippinen oder nach Thailand. www.swissinfo.ch/eng/climate-...

Neues Jahr, neue Plattform: Ich trenne meine Arbeit als Rednerin vom Consulting. Warum? Weil meine Keynotes keine Bulletpoint-Listen sind. Sie haben ihren Wert in sich selbst. Und ich ende garantiert nie mit "bucht einen Call". Deshalb neu: Klartext mit Tiefgang auf dorotheabaur.ch

AI replaces free markets as our new savior if we believe its prophets (not profits as in the automated subtitles), I say. It's classic TINA rhetoric: there is no alternative (see: en.wikipedia.org/wiki/There_i...) Full episode of the AI Sustainability Podcast: www.youtube.com/watch?si=SeU...

Würde man KI "zum Mitarbeiterinnengespräch aufbieten, würde sie wohl sagen: 'Gut, also, mit der Arbeit an Welthunger und Klimaerwärmung habe ich noch nicht begonnen, aber schau mal: ein selbstgemachtes Tittenbild von Taylor Swift – geil, oder?!'" www.srf.ch/news/schweiz...

Someone on LinkedIn wrote about this: "The problem is not that LLMs hallucinate, or that it's wrong to use AI for research. The problem is failing to use it critically responsibly." that was the entire post Rest assured: The day you read this kind of platitudes from me, my account has been hacked!

West Midlands Police Chief admits that AI "provided incorrect evidence in a report that helped inform a decision to ban Maccabi Tel Aviv fans from a match against Aston Villa last year. The report referred to a match ... that did not exist" OMG. www.bbc.com/news/live/c3...

"This is a good result but it is only the very first step in what is needed to maintain trust in Google’s health-related search results" You can't maintain trust that was never merited & never will be merited, as the flaws are fundamental, not merely cosmetic. www.theguardian.com/technology/2...

Sehr lesenswert: "Zwei Jahrzehnte vor dem rechtsblauen Mainstream hat Peter Handke einen fundamentalen Antagonismus erspürt, der im Kern darin besteht, den Fakten den Krieg zu erklären und natürlich den Medien, den 'Journalistenratten', auf medientaugliche Weise." www.faz.net/aktuell/feui...

The rise of genAI "crystallises Big Tech’s cultural takeover. AI turns reading and writing into automated industrial routines, optimised for speed and efficiency. It instrumentalises intellect, allows the... creative work of the mind to be outsourced to machinery" N. Carr unherd.com/2026/01/how-...

I've posted a little rant on LinkedIn about people who post screenshots without sources. I find it very annoying - across all platforms, i.e.: also here. Why don't you credit authors or websites? Why put the burden on readers to do the research? www.linkedin.com/posts/doroth...

Endlich Schnee. Kanonenfrei erzeugt, also quasi Bio, und Freiland. Vielleicht sogar Demeter - je nach Mondzyklus halt.

If I was a journal editor, the rule would be: Submissions with fake citations get an immediate desk reject, and all authors are banned from submitting to the journal for at least one year. www.rollingstone.com/culture/cult...

Must read! AI uses water like other industries - but golf course owners aren't dining with Trump, promising to save the world while destroying our jobs, keeping coal plants open, or threatening economic collapse. Big Golf isn't manipulating society. Big AI is. www.wired.com/story/karen-...

Mein Jahr 2025? Erfolgreich erfolglos. Ich wusste, dass Nachhaltigkeit bachab geht. Aber ich habe es irgendwann tatsächlich geschafft, gelassen damit umzugehen (ein Vorteil des Älterwerdens) und stattdessen neue Projekte anzuzetteln. Einen Rückblick gibt's hier. mailchi.mp/d5e6bfbb2b1e...

Das Bafu @bafu.admin.ch hat mich interviewt zu KI, Ethik und Nachhaltigkeit. Auszug: "Die Nutzenden haben weder die Verantwortung noch die Macht, strukturelle Probleme der Technologie zu lösen. Das ist Aufgabe von Politik und Unternehmen". Ganzes Interview hier: dieumwelt.ch/artikel/ki-p...

Participation in cultural creation does not begin with access to tools created from digitally looted art, but with respect for the work of others. It requires consent, attribution and compensation. Anything else remains cultural cannibalism in the name of innovation. (from my essay; translated)

L'IA générative crée textes, images et musique en quelques secondes, mais consomme beaucoup d'énergie et pose des enjeux éthiques. En recherche, elle est utilisée de maniere ciblée et sobre, p.ex. pour analyser des données environnementales. @dorotheabaur.bsky.social 👉www.lenvironnement.c...

Generative KI kann 📕, 🖼️ & 🎶in Sekunden erzeugen – doch sie verbraucht viel Energie & wirft ethische Fragen auf. Forschende nutzen KI gezielt & ressourcenschonend, etwa für Umweltdaten. «Man darf also nicht Äpfel mit Birnen vergleichen.» – @dorotheabaur.bsky.social 👉www.dieumwelt.ch/art...

Teilhabe am Kulturschaffen beginnt nicht mit Zugang zu Werkzeugen, die aus digitaler Raubkunst geschaffen wurden, sondern mit Respekt vor der Arbeit anderer. Sie verlangt nach Zustimmung, Namensnennung und Entschädigung. Alles andere bleibt ein kultureller Kannibalismus im Namen der Innovation. ⬇️

"human thinking is largely independent of human language — and we have little reason to believe ever more sophisticated modeling of language will create a form of intelligence that meets or surpasses our own" great piece by @benjaminjriley.bsky.social www.theverge.com/ai-artificia...

Researchers are "looking at whether AI facial analysis can give lenders useful clues about a person’s propensity to repay loans. For people without access to credit, that could be a blessing." Coming soon: cosmetic surgery that improves your credit rating! www.economist.com/business/202...

The big question for me is: Once the AI bubble burst, will these data centers become stranded assets or are there ways to repurpose them for "meaningful" uses of AI? Or does this require different types of chips/hardware? Can anyone help me with this?

"machines are supplying the language with which patients relay their suffering, and doctors are using machines to...respond". By outsourcing "the most intimate definitions of ourselves to AI, doctors and patients risk becoming...further alienated from one another" www.theguardian.com/us-news/ng-i...

"LLM’s output is bullshit in the philosophical sense: statements that are divorced from whether or not they are true or false.⁠⁠ ⁠⁠The point of a LLM is to output text that looks right given the training data and query, not to produce text that is right" pivot-to-ai.com/2025/11/07/t...

“It has been one minute, and you really cannot say anything?... If you don’t remember what you wrote, you don’t feel ownership. Do you even care?” Nataliya Kosmyna, MIT Media Lab research scientist on the effects of ChatGPT on writing and retention. www.nytimes.com/2025/11/06/t...

Wären alle Urheber:innen fair entschädigt worden, gäbe es generative KI in ihrer heutigen Form nicht. Die fossile Wirtschaft externalisiert die Kosten für CO2. Generative KI die für unsere Kreativität. Meine Aussage im Referat zu "KI & Nachhaltigkeit" im Rahmen der Agenda 2030 in Bern vorgestern.

It used to be the internet age and now it’s the AI age. "It’s normal to have something presented to you and be told that this thing is the future." What’s different is that in contrast to the internet, "AGI doesn’t exist". @shannonvallor.bsky.social www.technologyreview.com/2025/10/30/1...

Neuer Blogartikel: Woran KI-Übersetzung scheitert. Z.B. am fehlenden Savoir Vivre, im wahren Sinne des Wortes. baurconsulting.ch/camembert-od...

"Only Anthropic has built a tool that lets its models end conversations completely. But it’s for cases where users supposedly “harm” the model—Anthropic has explored whether AI models are conscious and therefore can suffer—by sending abusive messages" 🤦‍♀️ www.technologyreview.com/2025/10/21/1...

Recently I gave a keynote on "AI: what happens when we let machines speak for us - and about us?" at the FIT World Congress. In it I argued that automating translation means treating language as a matter of pure perception or prediction, at the expense of judgment. www.linkedin.com/posts/doroth...

"The surge in same-same AI-authored applications prompts employers to use robot filters to manage the flow. Everyone ends up in Tinderized job-search hell" www.theatlantic.com/ideas/archiv...

Kudos to the Oxford Dictionary of English for reading AI 1. as Amnesty International 2. as artificial insemination, 3. and only then as ... AI. 😁 (picture from Shannon Vallor's great book "AI Mirror - @shannonvallor.bsky.social 🙏)

"And attribution – or any other kind of recognition – is not just missing, it is actively prevented. Democratization would mean participation, visibility, and credit. What we are seeing here is something else entirely." 2/2 From @dorotheabaur.bsky.social

On the Myth of AI Democratization "This is not Robin Hood. It’s reverse Robin Hood. Value is extracted from underpaid authors, artists, and researchers in order to fuel systems that are controlled and monetized by some of the wealthiest and most powerful companies on the planet." 1/2

Newsletter. Heute dabei: - Warum KI-Unternehmen von Demokratie reden und Oligarchie meinen - Wie Sam Altman unlautere Forschung als Anlass nimmt, um seinen Iris-Scanner und eine Kryptowährung zu pitchen - Wie weiter nach dem Ende des Nachhaltigkeitshypes mailchi.mp/312b885babf3...

Talking about Nick Clegg apparently turns me into a kind of middle-aged Greta Thunberg: "How dare you?" But I stand by it: his justification for the genAI industry's predatory business model is the bottom of the barrel. #AI #genAI #copyright youtube.com/shorts/dgk4W...

My office for the next few weeks.

In the aftermath of the undercover study in which researchers from Zurich Uni tried to manipulate users, Reddit is considering using Sam Altman's Orb (an iris-scanning biometric surveillance tool linked to his cryptocurrency WorldCoin) in order to identify users. baurconsulting.ch/en/a-secret-...

Tech "built on top of the mess of society doesn’t just replicate its mistakes but reinforces them, because users think 'this ... can see everything" "therefore it has this view from nowhere'". But "there is no view from nowhere". @emilymbender.bsky.social www.ft.com/content/9029...

If a billion-dollar AI system can't "solve a problem that Herb Simon (one of the… godfathers of AI) solved with classical... AI techniques in 1957, the chances that models such as Claude or o3 are going to reach [AGI] seem truly remote" www.theguardian.com/commentisfre... @garymarcus.bsky.social