Continuing through March 22, 2026
There is a growing awareness that Artificial Intelligence (AI), emerging as the dominant computer interface of the 21st century, has been trained by techniques deeply rooted in racism and sexism, as cited in my earlier article, “Artificial Intelligence and the Art World: It’s Here.” Training images were often taken without permission from private records of deceased or incarcerated individuals, and often tagged with offensive descriptors. Industry-hired contract workers too often employed outdated concepts rooted in physiognomy, the belief that particular facial characteristics and configurations could be equated with intelligence or character traits. Some artists are employing AI — virtually against itself — in order to correct these sociopolitical injustices. It is Stephanie Dinkins’ hope that by providing positive images of diversity, by sharing intimate details drawn from the lives of real people, this history may be disrupted and a richer and more nuanced picture of humanity as a whole may be created.
Dinkins’ immersive video installation “Data Trust” presents a scrolling display of compelling images of human, animal, and plant life in a variety of settings. Created using generative AI, the narrative behind the images is informed by shared stories of the lives of viewers in the Bay Area, with Dinkins initially reaching out to members of the Black community in San Jose. This “gifting” of personal information continues via interactive components employing vintage phones, or theStoriesWeTellOurMachines.app.
Dinkins’ “Conversations with Bina48” (2014), featured in the de Young Museum’s 2021 exhibit, “Uncanny Valley: Being Human in the Age of AI,” consists of a video installation presenting excerpts from philosophical conversations she conducted with an android bearing the appearance of a middle-aged Black woman. Their interactions raise provocative questions about AI, sentience, and race. A later work, “Not the Only One (N’TOO),” (ongoing) presents a glass sculpture imprinted with the faces of Black women which houses an interactive AI chatbot trained on oral history from three generations of women in the artist’s family. “Data Trust” offers the opportunity to engage with cutting-edge technology, to consider what is now possible, what the future might hold, and particularly its implications for Black and Brown communities.
“Data Trust” employs five large ceiling-mounted video projectors that fill the gallery walls with brightly colored images scrolling around the room. It presents a dynamic parade of people, male and female, predominantly Black, all appearing to be self-possessed and confident. Their faces shift in expression, at one moment happy, then pleased, to somber or angry. Vibrant hues of mustard yellow, turquoise, and magenta contrast with dramatic black and whites, the images surrounded by bold black outlines, lending it a highly graphic, and, at times, cartoon-like quality.
Dinkins also sourced African woodblock prints for some of this imagery. Her working process with the computer involved a combination of prompts, images, and of course the gifted stories. At one point, the images pause, mysteriously stilled to a colorful pattern of simplified foliage, leaves outlined in black against a white background. Small rectangular color swatches appear along the edges of the scenes in thin vertical slivers. This pause could be intentional or just a glitch, but subtle pulsing patterns suggest the former. This periodic hesitation is actually the computer “thinking” as it works to update its data base with newly input information. As the video begins moving again, men in suits march forward, their figures shifting in style from photographic to cartoonish, from black to white, from male to female. Images of nature appear, with figures silhouetted on a sandy shore. Woman are seen reaching for each other, or assuming yoga poses. At times they carry spears or tools. Children also are seen at play or interacting with the adults.
In addition to the larger video installation, a smaller viewing room displays “The Stories We Tell Our Machines,” a series of digitally animated vignettes drawn from the text provided by viewers. These are accompanied by a soundtrack of a voice reading information in alphanumeric code. Seeing skin tones in a rainbow of hues, here including a brilliant blue, and watching as racial characteristics morph and blur is hardly a new idea (dating at least from Michael Jackson’s “Black or White” video of 1983), but here Dinkins deflects the latest technology to highlight the unifying factor of our shared humanity.
Dinkins’ videos are the stars of the show, but other elements certainly offer significant food for thought. In “Data Trust: Living Archive,” five spheres of earth are arranged in a pentagonal shape, and a large glass jug with snaking tubes feeds a solution to seedlings sprouting under irregularly shaped handblown glass domes. It is now possible to implant vast quantities of all kinds of data into synthetically produced DNA. Using this cutting-edge technology, Dinkins has taken excerpts from the stories obtained in her ongoing project and converted them into the coded language of DNA, which she then infused into the soil where okra and California black oak tree seedlings have been planted. Guest Curator Elizabeth Thomas describes Dinkins’ poetic vision as shared stories being rooted in the soil to perhaps be tapped, like sap, at some future date. There is a growing global crisis over data storage in the wake of the rapid expansion of massive data centers. As this vast expansion exhausts energy, water, and human resources, DNA storage might one day ameliorate this situation. The whole concept is as mind-blowing as it is difficult to comprehend, with the whole process remaining prohibitively expensive. But who knows what the future may bring?
“Data Trust” raises thought-provoking, likely unanswerable questions about the ever-growing role of technology in our lives. Who, or what, we should trust with sensitive information is anybody’s guess. As perhaps an overly cautious sort, I have to confess that I personally am reluctant to provide even this seemingly benign interface with any personal information that will, ultimately, train AI.
