AI, Ethics, and the Role of the Artist
Anna Ridler, Mosaic Virus, 2019, 3-screen GAN video installation, 30mins.
Using artificial intelligence systems as an artistic subject or medium comes with ethical implications and responsibilities. By SJ Bennett.
Artificial Intelligence (AI)1 allows artists to explore ideas and push boundaries in their work, and to collaborate in new ways with both people and algorithms. AI provides a medium for artistic work, and likewise, art can be a conduit for AI activism, with these two aspects often present within the same project. An illustration of this is found in art projects which make the social implications of AI tangible, exemplified by The New Real.
Thus, in behaving both as a mirror and a material, a human-machine amalgam2 and a tool, AI can provide a catalyst for reimagining the worlds we construct. But AI has its risks, too. AI systems are often anthropomorphized (seen as having human characteristics) and their inbuilt abilities overestimated, while their potential to reinforce existing societal problems is overlooked. When used for decision-making processes in sensitive areas such as policing3, schooling4 and healthcare5, for example, they can even result in discrimination against marginalised groups.6
How then should artists deal with these opposing possibilities, producing engaging art that satisfies their own creative intentions and social conscience, while being mindful of the ethical implications of their works, and without adding to AI’s mystification? What responsibility do artists have in contributing to their audience’s understanding of this technology? How can we understand the role of the artist in society, when for some it is an explicit intention and goal of their work to generate reflection on the implications of AI, but for others it is not?
Part of AI’s appeal for artists is that it enables exploration of the discrepancies between the real and the simulated. Artistic practices working with AI are often fluid and driven by the idiosyncrasies of the technologies, experienced as a symbiotic relationship; the AI output generates new inspiration for the artist and allows them to move beyond the binary to explore the shades in-between, materialising the unknown. The opportunities afforded by these characteristics position AI as an attractive medium. This can make its mystification more tempting when the artist is making art that acts as a framework upon which the audience projects its own interpretation.
In creating works to be experienced by the public, we can see that artists contribute to wider perceptions and expectations of AI. Yet artists tend to avoid prescriptive explanations. In doing so, they navigate the tensions (often subconscious) of balancing freedom to create engaging narratives with the unpredictability of audience reactions, and the artist’s ethical responsibility to that audience. We can better understand this balancing act by reflecting on the findings of Kate Crawford and Luke Stark, regarding their discussions with AI artists.
Borrowing from Simone De Beauvoir, Crawford and Stark7 describe tensions in creating AI art as forming the “ethics of ambiguity”; the notion that, although the outcomes of an act might be unknowable, the artist is still obligated to consider them to the best of their ability, and have a situated awareness of their role in their work and the implications of their actions. Crawford and Stark describe how the artists they interviewed were very conscious of the thread of ambiguity running through their works, from construction, to the material presentation and spatial location, to the experiences and knowledge audience members bring to interpretations and the moral implications of this.
This places a certain burden on artists’ shoulders. It is a responsibility of a different nature to that of, say, designers or technology companies who use AI to create tools for our daily lives. Artists working with AI need to recognise and respond to these responsibilities, but perhaps we as an audience can share this burden by exploring and understanding the contexts in which AI artworks are created. By doing so, we can hope for the activist intent in these works to be better realised, and for artists and audiences to have greater potential to contribute to the way technology shapes society and the way we live our lives.
1. For more about AI: https://www.coe.int/en/web/artificial-intelligence/what-is-ai
2. https://www.interaction-design.org/literature/book/the-encyclopedia-of-human-computer-interaction-2nd-ed/socio-technical-system-design
3. Green, B. and Chen, Y., 2019. The principles and limits of algorithm-in-the-loop decision making. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), pp.1-24.
4. [https://www.theverge.com/2020/8/17/21372045/uk-a-level-results-algorithm-biased-coronavirus-covid-19-pandemic-university-applications]
5. Nelson, G.S., 2019. Bias in artificial intelligence. North Carolina medical journal, 80(4), pp.220-222.
6. Buolamwini, J. and Gebru, T., 2018, January. Gender shades: Intersectional accuracy disparities in commercial gender classification. In Conference on fairness, accountability and transparency (pp. 77-91).
7. Stark, L. and Crawford, K., 2019. The Work of Art in the Age of Artificial Intelligence: What Artists Can Teach Us About the Ethics of Data Practice. Surveillance & Society, 17(3/4), pp.442-455.
