Author ORCID Identifier
0000-0003-1603-1357
Document Type
Other
Disciplines
6.4 ART
Abstract
Latent Space is a research-based artwork exploring the creative boundaries of co-creation with generative artificial intelligence. The work involved training a deep neural network and working within the latent space of the neural network to generate images that revealed the processes, assumptions, and biases immanent to generative AI.
Created for the Arts Council COVID Response initiative, the multi-faceted project trained a neural network on 3500 landscape photographs taken during six months of COVID lockdown.
While AI are noted for their ability to generate photo-realistic images, these photographs documenting walks within the limited range of lockdown regulations, were too few in number and diverse in content to achieve perfect renditions of landscapes that never existed.
Instead, the trained AI works within the gaps and omissions to produce evocative, abstract, and painterly approximations; landscapes that emerge from deep within the latent space of the neural network. This is the deep dream of the lockdown landscape: a series of videos and images generated by the neural network, where a process of machinic abstraction hallucinates the constrained and incomplete experience of lockdown.
The work is part of an ongoing series of works that explore co-creation with generative AI and the limits of machinic creativity.
DOI
https://doi.org/10.21427/xdgh-ga74
Recommended Citation
McGarrigle, Conor Dr., "Latent Space" (2020). Practice Based Research. 1.
https://arrow.tudublin.ie/aaschadppbr/1
Funder
Arts Council of Ireland
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License.
Publication Details
NetArt work published at conormcgarrigle.com/latentspace