A Symbiotic Future: Art - Hci - Ai
Date
Authors
Advisors
Journal Title
Journal ISSN
ISSN
DOI
Volume Title
Publisher
Type
Peer reviewed
Abstract
The emergence of new artificial intelligence systems, especially the advances in the fields of deep learning and computer vision techniques, bring not only technical challenges, but, and more importantly, new questions to the realm of art, and can modify the way in which we creatively use computer systems to produce language.
Even though the use of computers and artificial intelligence techniques is not a novelty in the art world – Harold Cohen (1928-2016), perhaps the most famous and relevant example, worked on his AARON program for 43 years –, the new scenario, bursting with supervised and unsupervised learning techniques makes it possible for algorithms to have sufficient autonomous decision-making power to choose paths that hitherto were only available to programmers and artists.
Cohen and AARON actively collaborated with one another, and in later years the program provided lines, basic abstract compositions, for the artist to colour in the spaces with colours, taking the program’s input as a starting point. Even in the software’s earlier versions, which could make both abstract and figurative sophisticated pictures, Cohen had the final say on what was or wasn’t aesthetically good.
The artist devoted a substantial amount of time to teaching AARON to draw and colour in, and hard-coded the decisions that the system could make. Currently, a growing set of libraries has eliminated the need to manually code the neural networks that provide the logic for systems capable of learning semi-autonomously by analysing a large amount of data.
Thus, we can delegate tasks to increasingly intelligent autonomous systems, including playful and aesthetic tasks. Given these circumstances, this paper reflects on notions such as intelligence, creativity and the role of aesthetics in a future where our relationship with intelligent artificial systems will become one of increasing symbiosis.
This initial more theoretical discussion is mainly developed in the light of the semiotic and philosophical theories of Charles S Peirce about the concepts of intelligence and abduction, the kind of thinking that, for the philosopher, is the foundation of creative processes. The technical advances in the field of algorithms employed in machine and deep learning are also considered. The final part of the paper analyses two works produced by the text’s author, establishing links between the previous discussion and the creative models of improvisation used to develop both works.
The first, “Hatred Apparatus”, is a system based on machine learning that autonomously collects comments from readers on websites, classifies them according to the level of hate speech they contain and displays them out of their original contexts. The second, “Apparatus Memories” is based on a deep learning system which carries out image segmentation operations on videos, separating the various elements that constitute the individual frames to assemble new autonomous narratives.