Enclosing or Democratising the AI Artwork World
- Robin Mansell
- Jun 22, 2021
- 16 min read
Updated: Mar 30
Introduction
Artificial intelligence (AI)-enabled prediction algorithms create multiple challenges to existing ideas about human agency and how the results of this agency may be governed. Weak or absent transparency in the operation of computational systems is changing the meaning of individual autonomy as AI enables vast numbers of new capabilities previously designed and implemented by humans.[1] The prevailing wisdom is that AI innovation is best driven by commercial market incentives. Investment in refining AI-inspired commercial strategies and techniques that are less and less susceptible to external (and even internal) control or oversight is central to futuristic visions of data-enabled societies. Among the many sectors entangled with AI innovation is the art world.
Hodge SCJ defines AIs as ‘computer systems able to perform tasks which traditionally have required human intelligence or tasks whose completion is beyond human intelligence’.[2] Computational technologies having this ability include machine learning, neural networks and predictive algorithms. When employed to create artefacts perceived as art, the resulting AI-assisted and AI-generated artworks are viewed either as a destabilising threat to the traditional art world or as an opening up of opportunities for new forms of expression. At present, AI art is principally the domain of computer science expertise and its AI component is mainly being driven by incentives in the commercial marketplace.
The agency to produce AI art has been harnessed to a commercial yoke. Is this an inevitable or desirable state of affairs? This paper examines the scope for ensuring that the expansion of AI in the art world does not lead to the enclosure of all these new forms of artworks in the commercial realm. It explores whether and how digitisation and computational advance can help to democratise art, opening rather than enclosing the artistic commons.[3]
The (short) commercial history of AI art
Google used its DeepDream neural network to classify artworks in 2015 and observed the potential for this AI system to be used to remix visual images. When the system was shared feely with artists, experimentation began. This led to a gallery showing of DeepDream-inspired artworks in 2016 at Gray Area, a San Francisco gallery and arts foundation. Artbreeder followed soon after as an open collaborative platform, with users making some 127 million AI-generated works at this writing.