Images generated by AI are now everywhere as people produce seemingly-high-quality artwork with just a few text prompts, be it for social media, company logos or picture books.

As AI hits art forms such as illustration, there are international artists looking beyond the relationship between AI and their artistic practice, far more interested in the impact of AI on culture and politics.

In fact, long before AI became a buzzword, these artists were using AI tools to dissect the technology, study how the tools worked and highlight their inherent issues. One such artist is Holly Herndon. Along with her partner Mathew Dryhurst, she was in September this year named by Time magazine as one of the 100 Most Influential People in AI for pioneering work in machine learning, software and music.

Their ongoing year-long research project, in collaboration with the Arts Technologies team of Serpentine Galleries in London, will be exhibited next year. In the words of Herndon and Dryhurst, the project explores the “dark corridors of what it means to be an artist in the AI age”.

According to Serpentine’s Arts Technologies curator Eva Jager, “The duo have been influential in the Berlin art and tech scene for many years now. But primarily, Holly and Mat are recognised as musicians and we are trying to position them as artists with a very pioneering practice that touches on music, mechanism design, Web3 innovations and more.”

Herndon and Dryhurst are not using AI just as a tool for “style transfer” or to make new sounds based on their past works. “They’re asking a lot of questions about ownership, identity and agency when they create new works,” said Jager.

This is particularly significant since just this year, AI was trained to create a song in the style of stars Drake and The Weeknd. The video, which garnered over 9 million views, was removed from TikTok, Spotify and other platforms due to claims by the artists’ record label Universal Music Group. The incident triggered a hot debate on copyright issues and foretold AI’s impact on the music industry. This theme is also explored in Holly+, known as Herndon’s digital twin, which was on show from Aug 19 to Sept 24 this year at Singapore’s ArtScience Museum’s Notes From The Ether: From NFTs To AI.

Holly+ is an AI-driven voice tool that can take any audio input and reinterpret it in Herndon’s voice, showcasing the creative possibilities and copyright challenges of AI voice models. Before the pandemic, the Training Humans exhibition at the Milan’s Osservatorio Fondazione Prada in Italy showcased photographs used by scientists to train AI systems to “see” and categorise the world.

American artist, geographer and author Trevor Paglen, a major name in international contemporary art and AI, roped in AI researcher Kate Crawford, author of Atlas Of AI, for the project.

Paglen told Prestige, “We were looking at representations of humans in the history of technologies, such as facial and object recognition, the relationship between computer vision and pseudoscientific fields, and the development in policing, military and commercial applications.”

The rise of social media gave those working on AI plenty of unconsented images to be used as training data. The images of people were often classified and tagged in labs by race, gender, age, emotion and personal trait, increasing the chances of reinforcing unwelcome societal biases. The exhibition by Paglen and Crawford aimed to highlight this gap in training AI.

Paglen is currently exploring the concept of PSYOPS (Psychological Operations) in relation to AI. He is interested in how, with the advent of generative AI, images and texts are increasingly generated specifically to sway our perspectives as intended by the authors. “I’ve been going back and looking at how intelligence agencies used things like neuropsychology and even UFOs as instruments of perception management,” the artist said.

British artist Christopher Kulendran Thomas explores similar interests in his art. Being Human, an immersive video installation created in 2019, is on show till Feb 25 next year, at the Singapore Art Museum’s Proof Of Personhood: Identity And Authenticity In The Face Of Artificial Intelligence.

Using a host of AI and machine learning tools, such as deepfaked recognisable characters like Taylor Swift to algorithmically synthesised pop music to auto-edited scenes scraped from the Internet, Thomas takes viewers on a journey from the fall-out of the violence of the war in Sri Lanka to the biennial of contemporary art founded in its aftermath, addressing human rights and contemporary art through the prism of AI.

“AI and machine learning tools give us one way of thinking about ‘being’ beyond the fiction…that we are individually and ontologically distinct from everything that is not us, the fiction that makes everything goes round,” said Thomas.

Intent and purpose notwithstanding, the use of AI in art looks set to increase. Serpentine Galleries’ Jager has spent years attempting to understand how artists are working with AI tools and how it is changing their practice. “Artists used to need specific coding abilities, they needed to get very technical with the technology. Since last year, we have have seen many interfaces where you don’t need to know how to code at all, so there’s been this proliferation of AI in artistic practices, also in the tools artists use,” she said.

Nonetheless, the future of AI and its impact on arts remains an open-ended question for now. Some globally renowned curators, artists and writers hope to see more art that look into the hidden and unintentional in AI, such as hallucinations and errors.

Yet, it is clear the most potent and memorable artworks that have come out of this field boldly excavate how AI seems to increasingly reinforce existing societal imbalances and concentrate power beyond the hands and minds of artists.

This feature was originally published by Singapore-based luxury magazine Prestige on 29 October 2023.