This project aims to visually uncover the subconscious of an AI by transferring all autonomy of decision back to the trained model.
With the emergence of diffusion models, the creation of photorealistic images has become a separate discipline. In order to artificially generate visuals, prompt designers create new grammatical syntaxes which act as an interface layer between the human and the machine. By taking away the prompt, the human becomes a mere observer of the machine originated output.
This project uses the open source model 'stable diffusion' by stabilityai and an empty prompt to allow us to observe the blackbox structure of a machine learning model. The images it generates, range from random patterns to partly identifiable scenes, and the timestamp marks the time the machine has imagined them since its creation.
an exploration by florian porada (2022)