Ghost in the Machine filmmaker Valerie Veatch aims to illuminate the impact of race science on today’s technological landscape.
Ghost in the Machine filmmaker Valerie Veatch aims to illuminate the impact of race science on today’s technological landscape.
Similar to many individuals, director Valerie Veatch found herself captivated when OpenAI unveiled its Sora text-to-video generative AI model to the public in 2024. While she didn’t grasp the technology in its entirety, her curiosity about its capabilities led her to discover other artists forming online networks to showcase their novel AI works. The prospect of engaging with others enticed Veatch into the realm of AI, yet upon her arrival, she was taken aback by the prevalence of images generated that were rife with racism and sexism.
Veatch became increasingly disturbed by the lack of concern from her fellow AI-enthusiast peers regarding the machine they championed, which produced hateful and prejudiced content without explicit instructions. This peculiar reality distanced Veatch from her initial engagement with generative AI. Nonetheless, it motivated her to create Ghost in the Machine, a fresh documentary examining the technologies and ideologies that set the stage for generative AI’s reality.
Rather than concentrating on the possible (if unlikely) advantages to society that generative AI advocates insist are imminent, Ghost in the Machine delves into the technology’s past to elucidate why it functions as it does today. In a recent conversation with Veatch about the documentary, she expressed her desire to document the origins of generative AI to provide clarity on the intense cycle of industry promotion we are currently experiencing. First, however, she needed to cut through the intentional vagueness from AI companies regarding the entire idea.
“To use the term ‘artificial intelligence,’ we must grasp what the fuck that term signifies,” Veatch told me during a video conference. “The reality is, it conveys nothing; it’s a term used for marketing and has always been. It’s a completely misleading, foolish phrase that has developed its own cultural interpretation, and I believe it’s crucial to be explicit about the language we employ and the significance of those terms.”
As Ghost in the Machine consistently highlights, the term “artificial intelligence” was first introduced in 1956 by computer scientist John McCarthy in his effort to secure increased funding for his initiatives. However, the documentary presents the inception of the term as just one of numerous significant milestones on a timeline that genuinely begins in Victorian-era England with the emergence of eugenics. Besides being Charles Darwin’s relative, Francis Galton was the pioneer of eugenics — the racist and discredited notion that human beings can be enhanced through the systematic eradication of “inferior” (interpreted as non-white) races.
While Galton certainly made some valuable contributions to academia, Veatch pointed out during our interview that it’s vital not to downplay the reality that his deeply held racist views significantly influenced the social sciences of his time. Galton and his protégé Karl Pearson were not directly involved in the creation of early computational devices. Nevertheless, Galton’s foundational research on multidimensional modeling — a method he employed to assess the appeal of African and European women — influenced Pearson’s thinking as he developed statistical methodologies like logistic regression, a fundamental element of contemporary machine learning.
Galton Pearson contributed to the normalization of the notion that individuals of different races held inherently divergent and measurable traits. This type of racist ideology led Galton and his contemporaries to believe that human intelligence could be quantified and that human brains operated similarly to machines. That leap, according to Veatch, played a notable role in persuading the public of the fantastical concept of artificial intelligence.
“What genuinely took me by surprise during my initial exploration of all of this was how swiftly, when analyzing the issue of superintelligence as a documentarian or journalist, one collides with the diminishing doorframe of race science because it’s interwoven into this technology,” Veatch explained, emphasizing that these notions are “infused” with eugenic ideologies.
Instead of attempting to refute the concept that generative AI models yield hateful ideologies because they have been trained on such material (a notion commonly referred to as “GIGO” — garbage in, garbage out), Ghost in the Machine employs its historical critique to elucidate why the entities developing this technology appear so indifferent to dealing with its contemporary challenges. This historical backdrop aided Veatch in comprehending some of her own disturbing encounters with generative AI when she was experimenting with an early version of Sora in an artists’ Slack group. Veatch recalls the community being friendly and inviting until another participant — a woman of color — began expressing concerns about the manner in which the model consistently whitewashed her whenever she asked it to produce images influenced by photographs of herself.
“It preserved her braids and her attire, but she was framing herself in an art gallery, which the program perceived to be a ‘white space,’” Veatch clarified. “My response was ‘what the hell,’ and I attempted to explain to the community why this was a significant flaw within the software itself.” No one else in the group responded to her post. “This was a Slack channel where, typically, there are always numerous screaming koala emoji reactions attached to every message. Yet this time, it was silent.”
Veatch took the initiative to reach out to OpenAI directly to notify the company about “how racist, sexist, and misogynistic the outputs [she] was witnessing — outputs where women would begin developing extra breasts and twerking after just a couple of rounds of generating a scene.” Veatch expected OpenAI to consider this a significant flaw that needed rectification before promoting Sora for broader usage; however, the company dismissed her concerns.
“The response I received was essentially, ‘This is quite cringe-worthy to bring up; we can’t do anything to change it,’” Veatch recounted.
That experience ignited a desire in Veatch to explore why various forms of generative intelligence consistently exhibit such distasteful, problematic behaviors. Initially, she didn’t believe that discussions with authors of research papers on the technology could lead to an engaging documentary, but her perspective shifted as she began to draw a direct connection from Galton’s eugenic statistic work to modern generative AI companies.
The individuals featured in Ghost in the Machine — a combination of AI researchers, historians, and critical theorists — present a compelling argument that virtually every aspect of the AI industry has been deeply affected by its historical ties to scientific fields designed to uphold discriminatory perspectives. When I inquired whether Veatch had ever been interested in directly speaking with the leaders of the companies criticized in Ghost in the Machine , she chuckled. Securing that kind of access, she remarked, would necessitate her to engage in various ideological contortions and make compromises that would implicate her film in the harms caused by generative AI.
“There’s this idea, you know, that these individuals wouldn’t trust just anyone,” Veatch stated. “Yeah, no kidding, and I genuinely hope they wouldn’t trust me. I’m not seeking to include them in the film, and they already speak to the media extensively. Am I going to embrace Sam Altman on camera? Is that a truthful depiction of this technology? That’s propagandistic.”
Ghost in the Machine will be available for streaming via Kinema from March 26th to March 28th before airing on PBS at some point this autumn.
Correction, March 21st: An earlier version of this article incorrectly referred to “logical regression” when the correct term is “logistic regression.”
