Design as Diagram of the System

07 january 2025

“The diagram of the piece is the physical scenario of how the piece unfolds...”

It’s about a balance of creating a set of parameters around a situation that allows for an exciting amount of chance, but not so much that it’s out of bounds. - MSHR

Growing archive of Event Scores Happenings 60-70’s Chance Operations Performance
An Anthology of Chance Operations

Self-Cannibalizing AI
Promptology & Recipes: Sound

Ingredients: voice, pack of cards, telephone book, nature, time

Mapping

Fluxus Island in Décollage Ocean - Nam June Paik

Cartography of Generative AI - Estampa

A Map of Days - Grayson Perry

A visualization and walkthrough of the LLM algorithm that backs OpenAI's ChatGPT. Explore the algorithm down to every add & multiply, seeing the whole process in action. https://bbycroft.net/llm

A Framwork for Understanding Sources of Harm throughout the Machine Learning Life Cycle, 2021 - Harini Suresh, John. V. Guttag

Diagram for a new work
1) Low tech enhancing spatial experience
2) High tech digital interaction
3) Instruments that guide visitor

Writing algorithms that dictate how shapes, colors, and patterns evolve over time. By manipulating parameters, achieving dynamic and intricate visuals. Combining elements of randomness with structured design, allowing for unique outputs each time the code runs.

Tools for the ritual: Platform for Human Body Interfaces - MSHR
Instruments
Noiseboxes AI Decoders Embodied Agents Chance Operators API Actors Meaningmakers Tricksters Alternative controllers Sonic Arcade

Possible outcomes of condensed highly curated momentum
Automated Drawings Conditional Cyber Sculptures Moments Snapshot of a stream Growing Layered Reality

Strategies
No-prompting Feedback Systems

Biases in AI
Learning Bias Aggregation Bias Evaluation Bias Deployment Bias Historical Bias

History of seeing AI development as purely accidental

Feedback as an artistic strategy to explore latent space

Liniarity is accidental in AI development

Numerical value brings possibilities to calculate distances, correlations, relations and probabilities. Backbone of image generation models. Tokenize text and image.

Feedback strategies
Text 2 Image AI as sophisticated noise generators, they do not generate images. Xray images, medical field is where they stem from.. CLIP and Diffusion models.

Go to model preferences
Misuse is good use

Removing Guidence Scale - Pure Diffusion, purest representation of dataset
Image to Image variations without prompt
Autoencoder (encode into latent space and decode back again)
AI 2 AI (img 2 img) (text2img2text2img)
Upscale / Downscale
Laion Aesthetic Predictor

Openjourney | Are we entering a purple future?
UMAP Mapping Embedding Space

All content: Cyanne van den Houten 2025 ©
Do not share this website or any of the content without direct permission from:
Cyanne van den Houten, +31 (0) 645403217 / info@cyannevdh.nl