Artificially Intelligent Painting – or is it?

Recent attempts at teaching computers how to paint with AI and NLP (Natural Language Processing), seem to go beyond the straightforward imitation of human technical skills. Artists-Engineers such as Ahmed Elgammal & AICAN, OBVIOUS, Robbie Barratt, Tom White and Boredom Research, look to encapsulate creativity and the essence of art in flexible algorithms, using different data sets drawn from art’s history.

In developing the algorithms, such practitioners base their works on two seminal trajectories to art's definition that either emphasizes representation (mimesis) or expression. However, within these traditional approaches to art, art is conceived predominantly as a human activity: psychology, perception and contemplation. Such new AI advances are arguably destabilizing this ancient axiom.

This paper looks to offer a theoretical framework for examining 5 different case studies of AI painting, through the prism of cyber-feminist critiques of technology, firstly by Katherine Halyes ‘Condition of Virtuality’ (Hayles, 1997) and Donna Haraway’s cyborg manifesto’ (Haraway, 1984). Such critique is used to extend Deleuze and Guattari's notion of a ‘Body without Organs’. Here, the concern is for an understanding of expression beyond the human body and from the point of view of what 'machine-expression' may mean.

Analysed through the lens of Nelson Goodman’s definitions of digital and analogue media (Goodman, 1968), painting is an interesting example of a medium, that even when used through digital intelligence, it still seems to resist any direct translation into digital systems of symbols. As such it may expose new trajectories of re-thinking art in a post-human context.