what_kind_of_ai_do_we_want._generative_ai_and_normalization [Felix Stalder, Prof. DfA]

“Liberate the possible from the tyranny of the probable.” (Dan McQuillan, Goldsmith University)

About:

In this module, we will combine theoretical and technical/practical approaches to AI, trying to develop a critical perspective and some hands-on experience at the same time. Our starting point is to consider «Artificial Intelligence» (AI) as a historical-material practice, that is, shaped by the concrete conditions of its development and use.

In particular, we want to look at the issue of “normalization”, the tendency of generative AI to create very similar-looking content, drawn from a relatively narrow range of possibilities.

Thus, we will look at contemporary artistic practices, read key texts to contextualize the development of AI within techno-capitalism, and have with guests who introduce us to the technologies that make AI appear intelligent and discuss their own (artistic) practices.

Date, Times, Location: Mo 04.03. - Fr 08.03., 09:15 -17:00, Room ZT 6.K04

PAD for Notes and References

“Introduction to Machine Learning”:

Guest: Alexandre Puttick (Latent Spaces Reserch Projekt).

Understanding the Latent Space: Basic mathematical operations. Slides

Playing a Game: Semantle.com

Quick Intro
WordVec Explained. Toward Data Science (Jul 29, 2021)
Taylor Swift Programming Meme Source

The Issue of Normalization

Jenka Gurfinkel: AI and the American Smile. How AI misrepresents culture through a facial expression. Medium. Mar 27, 2023
Eric Salvaggio: The Hypothetical Image. October 29, 2023
Meyer, Roland. 2025. “‘Platform Realism’. AI Image Synthesis and the Rise of Generic Visual Content.” Transbordeur 9.

Sources of Normalization

Probability vs Possibility
Bias in Data
- Mediengruppe Bitnik (2017) State of Reference (→ Analysis of the LAION-5B Data Set Knowing Machines.)
- Joy Boulamwini AI, Ain't I a Woman? (2019)

Corporate Policy
OpenAI Guardrails
Avoiding controversy (in the US)
Reinforcement Learning

Content Moderation
- Andrew N. Wilson, Workers Leaving the Googleplex, 2011, 11:03
- Frank and Eva Mattes, Dark Continent, 2015 Episode One 4:50, http://jnhvqx5annuswmmh3mhfj4exvartelzftr4gdc5chucllqqljccg3fid.onion

‘It’s destroyed me completely’: Kenyan moderators decry toll of training of AI models Guardian, 2 Aug 2023
Espinosa, H. (2026). “Working for the Machine – Fragments of life and death of a content moderator“. In: M. Miceli, A. Dinika, K. Kauffman, C. Salim Wagner, and L. Sachenbacher (eds.) Data Workers‘ Inquiry. Creative Commons BY 4.0.

Gillespie, Tarleton. Custodians of the internet: platforms, content moderation, and the hidden decisions that shape social media. New Haven: Yale University Press, 2018.
Gray, Mary L., und Siddharth Suri. Ghost work: how to stop Silicon Valley from building a new global underclass. Boston: Houghton Mifflin Harcourt, 2019.

Morning

Colonialism / Post-colonialism

Historical Process (1492-1974)

https://brilliantmaps.com/colonialism-history

System of Dispossession

“Colonialism = Thingification” (Aimé Césaire, Discourse on Colonialism)

Digital Colonialism

Infrastructure Submarine Cable Map, Facebook User per country, Meta's Data Center Map
Extractivism. Vladan Joler. New Extractivism CALCULATING EMPIRES. A GENEALOGY OF TECHNOLOGY AND POWER, 1500-2025 Crawford / Joler, 2023/24
Bias in Data, Data Sets
- Adam Harvey, Exposing AI
- https://haveibeentrained.com
Applications

De-colonizing AI

11:30 - 13:00 Nora Al Bardi (via Zoom)

Afternoon

Denormalization: The strange case of LOAB

Ange Lavoipierre. A journey inside our unimaginable future. ABC News, November 25, 2022
Matthew Sparkes. Why do AIs keep creating nightmarish images of strange characters? New Scientist, September 23, 2022
Steph May Swanson: What I Learned from Loab: AI as a creative adversary. 37C3, 2023, 36 Min.

Negative Weighted Prompts

Non-Normal Speech

Experiments with animal voices https://latentspaces.zhdk.ch/birdbot/

Waldrapp Field Recordings https://xeno-canto.org/361510

Voice Cloning https://replicate.com/afiaka87/tortoise-tts

Morning:

80% happy:Computer and Emotions

Paul Eckmann Universal Facial Expressions

Anger
Disgust
Fear
Happiness
Sadness
Surprise
(neutral)

Monda Lisa's Smile

Wikipedia Emotion Classification

Content analysis vs. behavioral analysis (involuntary facial expressions, gait, typing, biometric data…)

Crawford, Kate. Atlas of AI: power, politics, and the planetary costs of artificial intelligence. New Haven: Yale University Press, 2021.Chapter 5: Affect

OBJECTIVE OR BIASED On the questionable use of Artificial Intelligence for job applications, 02.2021

Afternoon: 13:30 -16:30 RAUM: ZT 1.D07 Probebühne

Guest: Manuel Hendry & the angry chatbot

→ 16:30 Cory Arcangel's Presentation (see pad)

Morning

Individual Work on de-normalized generation.

Possibility of personal “mentorat”

Afternoon

One more game to play