What follows is a quick map, not a deep dive: Sustainability, Research, Apps, Books & Documentation, Seminars, and Transart — five areas this community has actually been building in, this year.

One thing ties all five together: in every single example, AI is essential — either as object of study (research on language models, on alignment, on what a model even is), or as a tool that materializes an idea (vibe-coding, agentic AI turning a weekend sketch into a working app), or, often, both at once.

Everything mentioned here is real, live, and linked — most of it from udk.ai itself.

Time has come to weave three major research threads of my career— AI sustainability, AI ethics and alignment, and machine didactics —into a unified tapestry.

Freshest: a paper submitted just days ago to Nature Machine Intelligence — "On moral role models and the Sustainable Development Goals: A pre-registered test of role-model prompting in small language models." 

Nota bene: in case You are interested in abilities of modern AIs to provide novel insights into certain mathematical / cryptographic enigma known as "Elliptic Curve Discrete Logarithm Problem", feel free to stop by for a chat at our table in the Foyer.

Verwaltung: seminar-log, pushed this very week — attendance and activity points, extracted from photographed seminar forms and a Moodle list, via an in-house AI gateway. It now saves hours, every week.

Content Management: Kastalia KMS — thanks to agentic AI, the knowledge-management & lecture-broadcasting system which have been the backbone of all Einführungveranstaltungen of this house between 2016 - 2022 has been completely refactored. 

Practically all my lectures (and most of their recordings), presentations and Open Educational Resources are archived in Kastalia, up to know cca 5000 knowledge units / folios.

Education — HoloSky: an app, vibe-coded, for sky observation with HoloLens 2. Developed & deployed by Benjamin Herrmann.

Matrix Chatbot  Walden3 just won a small little 🏆 Prize in a Silicon Valley hackathon. Inspired by Thoreau's and Skinner's Walden books, Walden 3 is a bot whose semantic memories stored in a decentralized planetary storage mechanism known as Walrus. He paves the way for his future iteration, the principal Teacher character of my Habilitationschrift Walden4 oder Grundlagen der Bildungspolitik.

Voice Mixer, a locally hosted communal voice-cloning system allowing one to combine voices as a painter combines colors.

The topic of upcoming 9th UdK2030 Future Day  is  Zukunfttstag pflanzt Bäume, "plants trees." 

AI generated four flyer drafts were made; "Jahresringe" (tree rings) — option B, shown here — has been chosen by the community.

In times of budgetary cuts, neither documentation of Zukunftstag 8 nor udk2030.digital could ever exist without support from AI.

Primer 0: Developmental and Corpus Psycholinguistics — my first ebook about how a toddler builds a language, told through the research literature, a father's diary of his multilingual daughter, and hitherto unpublished data-analyses of speech of hundreds of children from the Child Language Data Exchange System (CHILDES) corpora.

Short History of the Learning Machine, is a leaner, more recent (e.g. post-GPT) sibling of my 1st Studium Generale course I gave as Juniorprofessor at UdK entitled Prolegomena to Educatio Digitale.  

Curriculum implements the classical continental didactic method of explaining a complex concept (e.g. a 2048-dimensional latent space within which artificial Minds execute operations with symbols) by means of explaining the history of the concept: from Pythagoras, al-Khwarizmi & Leibniz to Ada Lovelace, Claude (;) Shannon, Alan Turing and beyond.

Memes, Mindworms, Memplexes &c. — half a century after Richard Dawkins coined the meme "meme", it's time for a renaissance of the fascinating theory about how ideas spread & culture evolves.

Teaser for next week: we'll ask an AI to build — and actually run — a computational multi-agent model of gossip spreading within a community made of four status groups.

If that sentence alone gave you a feeling, you're exactly who this session is for. Feel free to join (Wednesday 16:00 - 18:00 at this very place).

One invitation before you go: this year's UdK.AI Ethics series — Can AIs distinguish good from evil?

25.11 "Problem" · 2.12 "Discussion" · 9.12 "Solution" · Aula Medienhaus, 18:15–19:45.

Everything here: udk.ai.