Friday September 19 2025

Block 2 10am

Lori Emerson - Excavating the Possibility of a Geocentric AI from Early Cybernetics
This block focuses on a possible history of AI based on cybernetics.

In this paper I return to some of the earliest articulations of cybernetics (by Norbert Wiener, Claude Shannon, Stafford Beer, and Ross Ashby) to show how there was always the possibility for an understanding of feedback and recursion that was decidedly disinterested in abstract, disembodied notions of brains and thinking that led to the version of AI we are currently living through. Drawing on insights by Hannah Arendt from 1963 on “The Conquest of Space and the Stature of Man” as well as contemporary writing by Yuk Hui, I show how one might return to the origins of cybernetics to imagine an AI that does not aspire to superhuman intelligence so much as it is invested in a model of thinking that is firmly earthbound, limited, embodied, and relational.

Xiaowei Wang - Glimmering gold and technical bodies

In taking seriously the contention „diaspora is a place of knowledge making“, this paper examines the rise of three models of human bodies and bodily knowledge — the porous body, the technical body, and the dislocated body, through archaeological material on Chinese medicine in the American West during the 19th century. Through these models of the body, I want to draw distinct connections to how these forms of the body and their enframings of nature are understood can be understood as a kind of cybernetics.

Xavier Nueno - The Principle of Least Effort or who Drives Scientific Invention

George Kingsley Zipf dedicated his career to developing a science of his own making: statistical human ecology. At its core was a simple observation derived from word counting. In any given text, Zipf argued, the most frequent word appears twice as often as the second most frequent, three times as often as the third, and so on. From Homer to Joyce, from Native American languages and language learning to dreams and psychosis, Zipf found this pattern repeated - evidence, he believed, of a universal ‘principle of least effort.’

Zipf’s law, as it came to be known, echoed earlier findings by Vilfredo Pareto in economics, and Samuel Bradford and Alfred Lotka in bibliometrics. It seemed to confirm a natural order in which a small elite governed while the majority followed. Yet it was within librarianship - especially in the work of John Desmond Bernal and the Association of Scientific Workers - that this supposed natural law faced its most pointed critique. Drawing inspiration from Soviet advances in bibliography and scientific planning, these critics recast Zipf’s law not as a universal principle, but as a historically situated social construct.

Block 3 2pm

Philip Leitner - the good, the bad and the ugly
A block on the subject human-in-the loop, which loop, and who owns it…

attention is all you need¹ sounds a lot like the First Rule of Influencing. It’s like a passage that points right to the top of socialmedia-sociocraty. This landmark paper from 2017, in sync with what would seem, for now, to be the largest marketing campaing for AI by openAI is opening up a new chapter.

A digitized humanity is actively improving machines in mimicking human behavior and language. Once we deduct things like so-called inovation or progress some old critiques of the industrial revolution will seem applicable. On that note, artificial intelligence is not as new or shiny as it seems.

[1] Attention Is All You Need

rybn.org (Marika Dermineur, Kevin Bartoli)

Marika will present the project HUMAN COMPUTERS that investigates the relationships between computing and labor organisation. rybn.org/human_computers

Kevin will present the project HUMAN PERCEPTRON, which invited participants to manually compute all the steps of a Perceptron, the very first neural network, designed to identify classes of objects. human_computers/humanperceptron

Block 4 5pm

Andreas Rathmanner - Content inbreeding - when AI digests itself
A plea for digital diversity, resistant data sets and a culture beyond algorithmic self-digestion.

In the future, generative AI will increasingly shape our information spaces. Already today, millions of people are prompting AI models to generate texts, images, and videos for us. But the more content is produced by AI, the more this content becomes the new reality - it flows as training data into the next generations of models, which in turn generate texts, images and videos, which in turn generate texts, videos and images. This recursive feedback leads to a kind of cultural inbreeding — a creeping erosion of diversity, context, and originality. Instead of drawing on the living complexity of the physical world, models are increasingly fed by algorithmically generated imitations. The models virtually forget what our world or diversity looks like, as they only use themselves as a yardstick. Instead of making new things possible, they narrow the horizon: the same usable, optimized, standardized patterns are reinforced again and again - what is available replaces the diversity of our world.

How does our perception change in a world fed by AI models? Will typical AI artifacts become part of accepted visual language? How do image composition and structure change when prompts instead of cameras set the rules? What are the dangers of a visual monoculture characterized by a small number of models and data sets? And what strategies can preserve queer, subcultural, or experimental styles from disappearing into homogeneous datasets? And what would a counter-model look like — a data ecology that fosters curated diversity, critical data awareness, and lively interference?

Marek Tuszynski - Information Sickness

In his lecture he explores how emerging technologies are quietly but powerfully reshaping society, politics, and the stories we tell. Rather than simply celebrating digital innovation, this talk examines its more unsettling effects—like hyper-profiling, emotional targeting, and the rise of synthetic trust. UlUltimately, it invites reflection on how technology can change behaviour and emotions, while simultaneously driving the infantilization of public debate and the erosion of democratic foundations.

Oskar Beneder - Artificial, Intelligent, Unclear – A Reality Check with Copilot
How AI works in real life, what it (doesn’t) do, and why human oversight still matters.

AI isn’t magic – it’s a tool. Sometimes a powerful one, sometimes just a very fast parrot with internet access. In this talk, I’ll show how companies actually use AI today, what’s behind the buzzwords, and why “smart” doesn’t always mean wise. Using Microsoft’s Copilot as an example, I’ll walk through how you can build your own agent, guide its behavior or at least try to and explore what happens when AI starts feeding on data created by… AI. Welcome to the age of data inbreeding, where diversity collapses and algorithms end up quoting themselves. My goal is to demystify AI, break down misunderstandings, and show how we can use these systems consciously – as tools, not oracles.