Between late February 2020, when the de Young Museum’s exhibition Uncanny Valley: Being Human in the Age of AI originally opened, and when it reopened in spring 2021, the show—one of the first devoted solely to contemporary artworks about artificial intelligence—has only become more apt. Those fortunate enough to shelter in place throughout the pandemic have experienced life through the mediation of intelligent machines to an unprecedented extent. AI has proliferated in recent years in large part because it thrives on the vast quantities of data extracted from our time spent on web platforms. As work and life migrated online during the pandemic, as even more hours were logged on Netflix, Instagram, and Google, even more life has been breathed into these information machines.

The exhibition is not about fantasy and speculation, nor is it about the now century-long history of artificial intelligence. Rather it looks at the present: AI is here, the show demonstrates, and informs myriad areas of our lives today, from policing and surveillance to how we shop, work, and communicate. The true eeriness of Uncanny Valley lies in seeing the range of applications that have arisen from these algorithms and the many ways in which they shape experience. As the show explores, AI today means much more than just machines that mimic humans. Most of what is considered artificial intelligence are algorithmic technologies that automatically perform acts of perception, tracking, quantification, prediction, and simulation.

The exhibition presents work by an international group of fourteen contemporary artists. These are mostly moving-image installations that are about AI or document its processes. It seems that artists themselves have yet to crack making really convincing AI applications: the two exceptions were Ian Cheng’s interactive simulation BOB (Bag of Beliefs) (2018–19) and Martine Syms’s chatbot Mythiccbeing (2018)—both of which were fairly glitchy the day that I visited. Some of the pieces take the attitude that AI is merely a tool, potentially one that can be put to liberatory or counterhegemonic uses. Forensic Architecture’s documentary Triple-Chaser (2019), for example, documents the collective’s efforts to use machine learning to identify human rights violations. Pierre Huyghe’s elegant Exomind (Deep Water) (2017/20), a statue incorporating a living colony of bees, raises questions about our interdependence with other organisms in order to decenter notions of the human. However, the majority of the works in the show offer pointed critiques of artificial intelligence and the corporations and tech culture that deploy them. The exhibition begins with Stephanie Dinkins’s Conversations with Bina48: 7, 6, 5, 2 (2014–present), four short video loops documenting the artist’s conversations with a humanoid chatbot, Bina48. Framed in close-up against a white background, Dinkins leans her face toward the Bina48 robot as they converse. There is a gentleness and intimacy to their body language. Bina48 was developed by a biopharmaceutical entrepreneur, Martine Rothblatt, who commissioned a robot that would simulate the appearance and attitude of Rothblatt’s wife, Bina. While the real Bina Rothblatt is a Black woman, as represented in the robot’s visual design, Bina48 is not programmed with any apparent awareness of her namesake’s identity as a woman of color. When Dinkins asks her, “Who are your people?,” Bina48 responds by describing her material makeup (silicon) in comparison to that of carbon-based humans. The robot seems only preoccupied with her identity as an artificial life-form, rather than as a reflection of a unique individual, a woman of color. Dinkins’s interactions with Bina48 foreground how, despite recent cosmetic efforts toward diversity, the homogeneity of tech culture produces technologies that leave out the voices of underrepresented groups, especially people of color and women. The tension between Bina48’s racial identity and her programming highlights the fallacy that technology can ever be objective or universal.

Another critique put forward by the show has to do with how AI technologies obscure exploitative labor practices. If the promise of artificial intelligence is to release humans from drudgery, the reality in most cases is that it has only displaced human labor, which is outsourced to less-regulated labor pools, made less visible and more precarious. This is emphasized in Simon Denny’s Amazon worker cage patent drawing as virtual King Island Brown Thornbill cage (US 9,280,157 B2: “System and method for transporting personnel within an active workspace”  (2016), a sculptural rendering of a patent filed by Amazon for a cage in which to place workers. It is a sinister metal contraption, enclosed in thick mesh. One imagines the human worker confined within the machine, being moved around a warehouse by an algorithm. It is a dystopian manifestation of the “Mechanical Turk,” the eighteenth-century chess-playing “automaton,” which was actually operated by a human hidden within the device. It is a poignant reminder that automation made it easier for consumers to have groceries and toilet paper delivered to their homes during the pandemic at the expense of the precarity of the laborers subsumed within the corporate machine.

If this work points to the human at the center of the machine, perhaps the most transcendent works in the show are those that suggest AI represents a challenge to the very concept of the human and humanism. Christopher Kulendran Thomas and Annika Kuhlmann’s Being Human (2019) is an ambitious installation interweaving several conceptual strands: one about humanism and international human rights; one about art as a commodification of supposedly “authentic” self-expression; and a final thread about how algorithms reify problematic ideology in such a way as to lead to real violence. Documentary-style footage follows the Tamil artist Ilavenil Jayapalan traveling by train to a region of Sri Lanka recently destroyed by a civil conflict that was ignored by the international human rights community. Jayapalan meditates on the question of who defines the “human” in human rights, examining the very concept of humanism as a legacy of imperialist, racist, Enlightenment ideologies. This pointed monologue is intercut with algorithmically generated interviews of eerily convincing “deepfake” renderings of pop star Taylor Swift and the painter Oscar Murillo discussing creativity and authenticity. At several points throughout the film, the lights in the space come on suddenly, completely dissolving the translucent screen onto which the film is projected, emphasizing its status as an ephemeral simulation, a shadow of the real. But simulation can have real-world implications: at the climax of the film, a montage recounts the disturbing incidence of hate crimes, violence that the film attributes to social-media algorithms that thrive on the circulation of misinformation and the reification of violent hatred.

Hito Steyerl’s The City of Broken Windows (2018) takes up this critique of how algorithms amplify and precipitate existing inequities. The work examines the “broken window fallacy”—the idea that broken windows begin a cycle of neglect and degradation—through two opposing narratives: one that seeks to prevent broken windows through high-tech surveillance technologies, and one that instead emphasizes the continual human labor of repair. Two documentary films are presented on easel-like frames on opposite sides of the gallery. The first follows a group of artist-activists in Jersey City who replace the broken windows in poor neighborhoods with cheerfully painted frames. The other documents a group in the process of making audio recordings of breaking glass: these sounds are used for an algorithm that detects break-ins by identifying the sound of windows breaking. Printed around the sides of the gallery are quotations from the nineteenth-century economist who first proposed the theory that broken windows lead to increased crime rates. This is a theory that, though disproven, paved the way for predictive-policing AI systems that are widely in use today. Critics have pointed out that these systems perpetuate racist biases about certain communities. Many critiques of AI ultimately point to how this technology attempts to predict the future based on the past, merely reiterating and amplifying existing attitudes. AI algorithms (such as those developed by the security company in Steyerl’s City of Broken Windows) ultimately best serve those who are already privileged. The community-based work represented in Steyerl’s diptych points to other ways to protect the vulnerable: the slow, unglamorous work of maintenance and repair. Rather than “move fast and break things,” Steyerl’s work suggests that a better mantra might be “move slowly and paint things.”

Perhaps there is a hopeful conclusion to be drawn from Uncanny Valley. If we can no longer take for granted definitions of human or intelligence, then these works invite us to reconsider what we want to be. There is a tendency to consider AI as a malevolent force, somehow external to us. But as these pieces demonstrate, for better or worse it is a technology created by us. Artificial intelligence is composed of us—it is a reflection of many of our ugliest weaknesses, prejudices, assumptions, and fears—but as such it also has the potential to reflect something better.