Harald Klinke

Thrilled to share that my book Cultural Data Science – An Introduction to R is finally out! A journey through digital methods, visual culture & data analysis—made for everyone exploring the intersection of humanities and code. #DigitalHumanities #RStats link.springer.com/book/9783031...

Excited to share a new preprint, where I read generative world models against the intertwined history of machine vision and computer graphics and draw out how "human-human" visual culture is becoming operative in these models: doi.org/10.5281/zeno...

What changes when AI learns conversation directly from sound? Block’s Dessa was reportedly trained entirely on human speech. It listens and speaks simultaneously, learning timing, laughter and intonation alongside words. block.xyz/dessa #SpeechAI #MultimodalAI #HumanAIInteraction

A striking Foreign Affairs piece by Niall Ferguson, Lisa Klaassen & Eric Schmidt: AI is becoming not just a technology stack, but a power stack: chips, energy, compute, capital, models, standards. For Europe, sovereignty cannot stop at “European models.” #AISovereignty #AIInfrastructure #Geopolitics

AI sovereignty is not autarky—it is the capacity to act. The UK’s CST asks: what must we own, co-develop, or access securely across compute, data, models and skills? #AISovereignty

Heute startet die re:publica x New Fall Festival in Düsseldorf! Am 9. und 10. Oktober diskutieren Expertinnen und Experten im NRW Forum über KI, digitale Gesellschaft, Politik und Kultur. Der Eintritt ist frei. #rpDUS #KünstlicheIntelligenz #DigitaleGesellschaft

Having long been aware of those tech bros who read my work as utopian porn, where they’ve taken that doesn’t exactly surprise me.

Was bedeutet digitale Souveränität, wenn vollständige Unabhängigkeit nicht realistisch ist? In diesem Artikel schlage ich dafür den Begriff „Post-Lock-in“ vor: einen Zustand, in dem Abhängigkeiten transparent, beherrschbar und reversibel werden. #DigitalSovereignty www.linkedin.com/pulse/post-l...

A 217 year old Napoleonic cipher has been cracked with GPT-6 Astra. A striking example of how AI can turn previously uneconomical research problems into feasible #DigitalHumanities projects. #DigitalHumanities #AIResearch #ComputationalHistory

@harald-klinke.de counter proposal: identify valid research questions that are not feasible for LLMs to solve.

Our core argument is that LLMs should not be regarded as reliable simulators of human participants because subtle changes to the input text can lead to very different reactions from humans and LLMs.

In heterogeneous system of AI agents, when agents disagree, who persuades whom? We found: 1) Models easily influence each other 2) Model size/certainty doesn't imply resistance to influence 3) Opinion shift is determined mostly by the resistance of the hearer. Preprint: arxiv.org/abs/2610.03095

OpenAI introduced Dots yesterday (Tue 29 Sept), joining Meta's Muse and xAI's Grok Bot. Deep geeks have been using similar tools with names like OpenClaw and Hermes, but the learning curve for those was quite steep. Dots, Muse, and Grok Bot, however, are insanely easy to use and scarily powerful.

How much of an AI coding agent’s performance comes from the model, and how much from its harness? A new study finds no clear overall solve rate advantage for vendor native harnesses, but substantial differences in cost, speed and tool use. #AIAgents #AICoding #LLM

The first DAHJ Conversations event is coming up! On Sept. 22, Steve Rossi, Sam Zipper & Makenzie Gallagher discuss 39.409001, -100.645781 — where digital art, environmental science & coding meet to make complex systems visible. #DigitalArtHistory #DigitalArt #EnvironmentalHumanities

Bitte RT: Europäische Behörden, Abgeordnete u. Verlage auf Bluesky: Ihr könnt euren Account zu Eurosky nach Europa umziehen. Zehn Minuten, kein Follower verloren, gleicher Handle. https://move.eurosky.tech

Mein #ikonographischesGedächtnis hat wieder einen seltsamen Bezug hergestellt:

Interesting piece on the fragmentation of academic social media: The key question may no longer be where “Academic Twitter” went, but how academic publics are being redistributed across platforms. #AcademicSocialMedia #Bluesky #HigherEducation

AI.Lab-Webinar: Lernen Sie die Open-Source-Plattform für KI kennen 👨‍💻🔍 Am 8. September, 10:00 Uhr, zeigt das AI.Lab Supportteam im kostenfreien Webinar, welche Möglichkeiten die Open-Source-Plattform für KMU und Start-ups bietet. ℹ️ www.ki.nrw/ai-lab-webin...

After scraping around 12 million artworks from Cara, the developer is now collaborating on Lantern, an open source tool helping artists track their work in public AI datasets. A fascinating turn in the debate over AI, provenance and control. #AIArt #Copyright #Provenance

Yeah mostly Aptroto is the name of the protocol, atmosphere is the name for all the things built on it Similar to how the web is build on a set of protocols but we call it “the internet” or “the web”

Sehr interessant: @hkohle.bsky.social zeigt, unbezahlte Mikroarbeit kann ausbeuterisch sein, Qualität ist nicht selbstverständlich und manches Crowdsourcing erzeugt genau die Daten, mit denen #KI menschliche Arbeit ersetzt. Trotzdem: Wie kann Öffentlichkeit an Wissensproduktion beteiligt werden?

Inherent is training Faraday, a research agent based on a 27B Qwen model, to reproduce published research, choose experiments and use coding agents as tools. Its Replica benchmark results suggest that specialized agent training could matter more than model size. #AIAgents #AIResearch #OpenModels

Brazil is pairing US and Chinese AI compute with RISC-V and a national cloud. Sovereignty means diversifying dependencies. Europe should notice. #AI #Sovereignty

Forschungsdatenmanagement braucht gute Orientierung. Maximilian Heber hat mich auf den Relaunch von forschungsdaten.info hingewiesen: klarere Navigation, fokussierte Texte, filterbarer Kalender und neue Stellenübersicht. #FDM

This is my curated feed of folks engaging critically with technology, particularly around AI. If you find it useful, please link and share with your networks! bsky.app/profile/did:...

Great team! And the database of The Vilhelm Hammershøi Digital Archive (ViHDA) is here (with sliders): hammershoi.smk.dk

Die Belvedere Summer School 2026 in Wien widmet sich den digitalen Strategien, Methoden und Tools von Sammlungsinstitutionen. Themen reichen von Digitalisierung, Linked Open Data und KI im Museumsbereich bis hin zu digitalem Storytelling und nachhaltigen digitalen Praktiken. Reise und Unterkunft […]

We will have a (in person) mini-conference at @mpilhlt.bsky.social in October on "Legal History Meets Digital Humanities: Digital Methods and AI at Work at the Institute". I'm really looking forward to it 😁 https://www.lhlt.mpg.de/events/46449/2077701 #LegalHistory #AI #DigitalHumanities

Interesting opportunity at GESIS: where secure data access, reproducible research, and legal compliance meet research infrastructure. #ResearchInfrastructure #TrustedResearchEnvironments #OpenScience

Umnutzen und strategische Unterlaufen von KI-Systemen kann auch Ausdruck neuer Handlungsmacht sein, nicht nur Anpassung an „prefabricated futures“.

Entscheidend wird nun weniger der Gesetzestext als die Praxis: unabhängige Audits, belastbare Tests und Transparenz, ohne Europa zum Weltmeister im Regulieren von Modellen zu machen, die anderswo entwickelt werden. #AIAct #AISafety #DigitaleSouveränität

Really interesting perspective on how digital cultural heritage depends not just on digitisation, but on sustainable infrastructures for access, preservation and discovery. #DigitalHumanities #CulturalHeritage #DigitalInfrastructure

WTAF? Hot take: if you want people to believe you have made progress towards AGI, look like you are investing in what comes next. But do it in the cheap and meaningless scale. https://det.social/@HxxxKxxx/114343347357269502

We also currently have one open position for RIDLOP - so please take a look and share with people who are into Trusted Research Environments and sensitive data: gesis.jobs.personio.de/job/2707934?... #TRE, #fivesafes, #secureaccess

🚀 Just published: DentroChat – European AI chat with complete data sovereignty GDPR-compliant AI chatbot running entirely on European infrastructure. Generate images, analyze files, search web with fast, thinking, creative modes. euroalternative.eu/dentrochat

Any news report that starts “OpenAI says…” or claims “an AI” (sic) has “broken containment” or “gone rogue” is no longer doing journalism, but PR for the #AI industry. #OpenAI used their product to hack a third-party and failed to implement adequate safeguards. That’s it. That’s the story. #AIHype