Searching for surprises in the uncanny valley
How do you judge the output of a generative AI?
How do you use that judgement to identify what's wrong, and what can be changed to build a better AI?
An open-space session where a group of us trains DreamBooth to look like people in the workshop, give the trained model lots of prompts, and collectively try to feel our way through judgement back to principles of how to train and trigger the AI.
Building on a workshop from James and Bart.
