Flow
2021 - 2022

Exploring the relationship between the photograph and AI, Flow has been created with the use of a generative adversarial network and a dataset of over a thousand photographs. This early development of the large generative AI models we see today is typically used to create images of one specific thing. A model that generates images of faces would have to be trained with photographs of faces, for example. However, this model is trained with a massively varied dataset, containing everything from microscopic images of cells to deep space images of nebulae. Trying to make sense of this uncorrelated data, the machine learning model synthesizes the abstract from the representational, highlighting the direct impact that training data has on the product of an AI algorithm.