Andreas Wagner

2 kinds of news stories about AI:

1) "What happens if <0.1% of people make the tools the rest of us use to think?" No buzz. It's vague & abstract.

2) "AI is alienating a generation of kids from their parents." Picture of Nåndõ, who let an LLM choose her name & shocking hairstyle. Wow! Viral. +

Readers like concrete stories with human interest — especially if they’re about sex, or wackos, or kids doing bad things.

The big problem of power and inequality is *too* big to make a good story. Also it’s not news. But imo amplifying concentration of power is the biggest risk of AI.

I know it's more about case law and deference between European and national courts than about legal standards embodied in Directives and Regulations, but would not the Solange II decision require in some way that when the EU dissolves the protections of GDPR and AIA, national rights protections […]

URGENT OPPORTUNITY – Deadline November 20!

The University of #Trento is looking for 10 students from conflict zones for one year of free university education!

✅ Full Foundation Year
💰 €7,200 #scholarship
🏠 Free accommodation and meals
📚 Italian courses + university preparation
🎫 Transportation […]

Wow, *two* big outages today, in the top web frontend (CloudFlare, in front of about 25% of sites), and the top git forge (gitub, with something like 97% of public git repositories).

Stuff like this is why I built https://arewedecentralizedyet.online/ to keep an eye on how centralized some of […]

#TCDXII Our twelfth conference dedicated to Caribbean digital studies. Jamaica, 2025.

And we're back! Join us December 4–5, 2025 at Northeastern University in Boston for our 12th annual The Caribbean Digital Conference. Our schedule is packed with wonders and amazing colleagues. Check it out! As always, attendance is free, whether zoom or live. Hope to see you there!

UW joins 'first wave' of colleges focused on AI The initiative, AI@UW, was made possible in part by a donation from Charles and Lisa Simonyi, a husband-wife pair of philanthropists linked to Microsoft.

New initiative at UW to be led by @nlpnoah.bsky.social. This comes on top of the NSF-NVIDIA $150mil investment in a national fully open AI ecosystem, led by @ai2.bsky.social and Noah. Good news for UW; imo probably good for academia more broadly.

RE: https://mastodon.social/@Gargron/115569820207257167

One of the things we don't talk about enough is how to secure the peaceful transition of power in open source projects.

We've all seen BDFLs turn into mad kings.
There are innumerable projects which have collapsed when the founder left […]

The new features are still in alpha—but @djevans.bsky.social's contextual tools for Viral Texts data are live clusters.viraltexts.org

Click "View witness in context" to see a given reprint on the newspaper page, alongside other reprints on that page—click a cluster ID here to see its other reprints

#OCLC has refiled its motion for default judgment against Anna's Archive: https://www.courtlistener.com/docket/68157923/57/oclc-online-computer-library-center-inc-v-annas-archive/

Europäische PayPal-Alternative: Bezahldienst Wero für Online-Einkäufe gestartet

Sparkassen, Volksbanken und etliche Händler sind schon dabei: Wero als europäische Antwort auf Dienste wie Paypal startet für Online-Shopping […]

Very persuasive. Worth underlining that there are trade-offs. If your goal is to "improve the writing," but also "keep this accessible to the widest possible audience, and don't advocate any specific perspective," you are pulling and pushing on the same lever. Humans can't both pull & push either.

I just read the testimony of these 60 former DOJ attorneys in the NYT. I couldn't look away. If you have too much stuff to do today, don't start reading. Here's the tl;dr, which you already knew anyway: DOJ is gone gone. It's just the private legal army of a cruel madman.

Mein Kommentar bei @riffreporter zum Berliner EU-Gipfel für digitale Souveränität: „Zu einem Gegenmodell sollte es gehören, die öffentlichen Bibliotheken durch erheblich mehr Ressourcen und Stellen für IT-Experten zu Knotenpunkten des europäischen Internets zu machen oder in sozialen Netzwerken […]

So if I'm reading this right: https://www.microsoft.com/en-us/security/blog/2025/11/07/whisper-leak-a-novel-side-channel-cyberattack-on-remote-language-models/

(If one forgives the sampling limitations (one question, 100 positive responses, etc.), which I'm not sure you should...)

The output […]

Staatzschutz ermittelt: Angriff auf Leiter deutsch-arabischer Schule in Neukölln Er setzt sich für einen arabisch-israelischen Austausch ein. Nun ermittelt die Polizei zu einem Angriff auf den Schulleiter in einem U-Bahnhof. Regierungschef Wegner ist erschüttert.

Dr. Hudhaifa Al-Mashhadani, Leiter der Deutsch-Arabischen Schule Neukölln, wurde am Freitag Opfer eines mutmaßlichen Mordanschlags. Ein Mann versuchte ihn am Bahnsteig vor eine einfahrende U-Bahn zu stoßen und zeigte danach eindeutige, drohende Gesten. 1/x

"Modern luxury is the ability to think clearly, sleep deeply, move slowly & live quietly in a society designed to prevent all four."

Infoflyer for the announced event. Left: Pictures of Newspaper articles. Right: Text with announcement. The text reads: "Humanites Numeriques Tourangelles, Mercredi 26 novembre 2025, 17h-19h - Tours, CESR - Salle Rapin et en visioconférence. #9 Argument Mining in News Media: Tailoring Models and Methods for Responsible Application, Sarah Oberbichler - Digital Humanities Lab, Leibniz Institute of European History."

Can we trust general-purpose LLMs for rigorous academic work? According to @SarahOberbichler from #DHlab_IEG, research integrity demands specialized AI models, not general-purpose LLMs when using AI as an analysis tool. Join her upcoming talk online or at CESR […]

[Original post on fedihum.org]

Ankündigung HERMES-Netzwerktreffen

Promovierende aufgepasst: Ihr seid euch unsicher, ob Topic Modeling das richtige für eure computergestützten Auswertungen ist? Oder doch lieber (nicht) Netzwerkanalyse? Und welche Alternativen gäbe es? Am 17. 11. (16-17:30 Uhr) habt Ihr die Gelegenheit, euch […]

[Original post on fedihum.org]

Just received the great news that the video recordings from #DH2025 have been published on the conference website https://dh2025.adho.org/. But it would be great if these were published with an explicit license and a unique URL (or DOI) for each recording to allow for re-use and distribution […]

Auch das gehört übrigens zum Themenfeld
#Generationengerechtigkeit #Rente
#JungeUnion

The diversity of metadata provided by users uploading items to @internetarchive caused me to (again) make the most common error in API calls: making them too specific. I had been too optimistic and limited my search to items explicitly to those marked as being in Arabic. Removing this condition […]

So the goal of this post is not to weigh costs and benefits, or argue with skeptics. It is purely to sharpen our sense of the potential upside latent in a vision of AI as “cultural technology.” I think one reason that phrase hasn’t cheered anyone up is that it has been perceived as a deflating move, not an inspiring one. The people disposed to be interested in AI mostly got hooked by a rather different story, about protean general intelligences closely analogous to human beings. If you tell AI enthusiasts “no, this is more like writing,” they tend to get as depressed as the skeptics. My goal here is to convince people who use AI that “this is like writing” is potentially exciting—and to specify some of the things we would need to do to make it exciting.

Eventually our fixation on individual "intelligence" will be displaced by a recognition that we're talking about collective knowledge and culture. Interventions on that scale are just as scary, but at least we'll understand the stakes better. tedunderwood.com/2025/07/02/a...

Did a fair bit of reviewing contributions to the #deRSE26 today. Lots of #AI contributions. A pity.

I have to admit that I am not /the/ expert in that field. But I would argue that ordinary data analysis should be drawn to RSE conferences these days. Have a track "AI & Data Analysis" draws too […]

A screenshot of Thimbleweed Park. Agent Ray stands by the Pigeon Brothers. There's a list of verbs on the bottom of the screen, they are arranged in a grid of 3 by 3 words.

I love oldschool #pointnclick #AdventureGames!

In theory. In practice I very quickly find myself so tired of them being so slow!

A few ideas for devs:

- Settings to: Change the speed of walking/talking, allow instant movement to adjacent scene when double […]

[Original post on tutoteket.no]

my husband made it so the home assistant voice assistant can use full-on ollama to interpret instructions. then he told it to turn the lights on. it's been explaining its thought process for over five minutes now and the lights are still off.

Eine Anleitung für einen Kartentausch zur Themenfindung.

Sehr gute Idee zur Themengenerierung bei einem #Barcamp - hier bei der #edunautika

Computer punch cards with a hole in them, in the window is a microfiche

I just learned about a new data storage format. A rare analog digital hybrid. Behold the "Aperture Card"

This is a punch card that holds a microfiche slide with a document, diagram or photo.

The punch card could be used to catalog the scanned document or to […]

[Original post on sauropods.win]

I’ve been testing a theory: many people who are high on #AI and #LLMs are just new to automation and don’t realize you can automate processes with simple programming, if/then conditions, and API calls with zero AI involved.

So far it’s been working!

Whenever I’ve been asked to make an AI flow […]

Die USA drängen auf direkten Zugriff auf EU-Polizei-Biometriedaten.
Fast alle EU-Staaten wollen ihr visafreies Reisen erhalten – doch der Preis wäre enorm: Zugang zu Fingerabdrücken, Gesichtsbildern und Identitätsdaten.
Im Rat gibt es massiven Streit: Einige Regierungen wollen klare Schranken […]