Earlier this year OpenAI released ChatGPT, which showcased the ability of AI to write text, tell stories, and code to a pretty decent degree.
It marks the latest development in a wave of generative AI technology, including Stable Diffusion and DALL-E2, which have proven capable of generating incredibly sophisticated images. While these technologies have gained a considerable amount of attention, AI has been curating what we see ever since Google first deployed its PageRank algorithm to provide a degree of order to the internet.
The notion of AI curation has gained notoriety in recent years as more and more of us have accessed information via social networks, with nary a platform avoiding accusations of helping to spread misinformation in some way, shape, or form. Indeed, the UK aims to make social media executives criminally liable should they breach their duty to keep children safe online.
Curating art
‘The Algorithmic Pedestal‘ offered a more traditional take on AI-driven curation. The project, which was created by Oxford University researcher Laura Herman saw the Instagram algorithm pitted against artist Fabienne Hess to curate art taken from the Metropolitan Museum of Art’s open-access collection.
The exhibition, which took place at the J/M Gallery in London, saw Hess’ selection displayed alongside that of the algorithm, which curated a selection of images from the collection to display on the homepage of the profile setup by the researchers.
After speaking to Hess, she explained that her selection was driven by a sense of loss, with each piece of art designed to reflect that in some way. Her collection, the Dataset of Loss, has been formed over a number of years and represents resistance to the algorithmic way of seeing that has come to dominate modern society. The “black box” nature of the Instagram algorithm means that no such intentions can be inferred from its selections.
“The absence of a clear pattern or priority in the algorithmically-curated images highlights the concerning nature of the “black box” algorithms: given the complete lack of transparency, we do not know why certain images are being selected,” Herman explains. “These algorithmic selections are deciding what people see, determining what they are exposed to, and driving their cultural reality—which most people do not realize.”
Dark arts
The act of curation is inherently one of choosing a small part of a collection to put on display, with most of what museums and galleries possess not on show. That curation process has come in for some criticism recently. For instance, the Wellcome Collection’s Medicine Man exhibition was said to “perpetuate a version of medical history that is based on racist, sexist and ableist theories”.
The exhibit, which featured a range of medical oddments, including false legs and Napolean’s toothbrush, was shut down shortly after the criticism emerged. Other museums across the capital have also come under fire for the supposed theft of artifacts and exhibits from other countries. The British Museum, for instance, is in seemingly perpetual talks about the return (or otherwise) of the Parthenon sculptures.
These controversies have meant that many exhibits are now accompanied by sorrowful descriptions that aim to demonstrate that the past was another time with different morals and societal norms. For instance, Tate Britain famously assigned a Hogarth painting with a label that pondered whether the legs of the chair Hogarth was sitting on might “stand in for all those unnamed black and brown people enabling the society that supports his vigorous creativity”.
Thinking for us
It’s generally not the job of museums and galleries to tell you what to think, but rather to make you think and evaluate that which is before you. In other domains, the application of AI has reduced our ability to do just this. Maps, for instance, have largely removed our need to use our own wits to navigate and provide us with turn-by-turn instructions. While London taxi drivers typically needed to take the hugely challenging “Knowledge” exam, a modern Uber driver can make do with the satnav on their phones.
Hermans believes that with the rise of AI-powered content creation, it will inevitably require AI-powered curation to enable us to make sense of it all. Whether this sparks a backlash against technologies that are increasingly doing the thinking for us or we’re happy to outsource our critical faculties to machines is something that remains to be seen.
“Even if they don’t take the drastic step of avoiding such feeds, they may approach them more critically and seek out other spaces to come across pieces more organically,” Hermans says. “Art, after all, has historically been about novelty and originality; statistical models grounded in pattern matching and similarity do not necessarily suit this field.”
Master or slave?
The nature of our relationship with AI is something that UCL’s Tomas Chamorro-Premuzic ponders in his recent book I, Human. He argues that in many instances, it’s not the mundane things that are being automated but our very thought processes themselves. We’re not being freed up to think more as technology is doing that for us.
“There is not much evidence that the rise of AI has been leveraged in some way to elevate our curiosity or intellectual development, or that we are becoming any wiser,” he explains. “Our lives seem not just predicted, but also dictated by AI.”
It makes perfect sense from a technological perspective for so much of our lives to be automated, but it makes far less sense from our own perspective. Much of our lives seem to be optimized for and by technology, and Chamorro-Premuzic suggests that we have far from become freed by automation and instead surrendered our humanity to it.
“Our very identity and existence have been collapsed to the categories machines use to understand and predict our behavior, our whole character reduced to the things AI predicts about us,” he continues.
Nowhere is this demonstration of humanity more evident than in the art we produce. If technology is both producing and curating art on our behalf, then that seems to miss its very point. “Robot” derives from the Czech word for slave, but whereas the protagonists in Capek’s famous play envisaged their robot creations serving them, the reverse is actually what transpired.
While we’re not in the midst of some Terminator-style battle for humanity, we should nonetheless strive to ensure that as AI develops, it does truly live up to its promise of liberating us from drudgery and freeing us to do those things that make us human. Just as with Capek’s creation, at the moment, it runs the very real risk of doing the exact opposite.
