The flavor of generative AI raises ethical concerns.

Director Valerie Veatch found herself captivated when OpenAI unveiled its Sora text-to-video generative AI model in 2024. Although she didn’t grasp all the technical details, her curiosity about its capabilities and the burgeoning online communities of artists sharing their AI creations drew her in. However, her excitement quickly turned to dismay as she encountered the technology’s tendency to produce images rife with racism and sexism.

What troubled Veatch even more was the indifference of her fellow AI enthusiasts towards the hateful and bigoted outputs generated by the model, often without any explicit prompts. This unsettling experience pushed her away from her initial explorations of generative AI but also inspired her to create *Ghost in the Machine*, a documentary that delves into the historical and theoretical foundations of generative AI.

Rather than focusing on the potential benefits touted by AI accelerationists, *Ghost in the Machine* examines the technology’s history to clarify why it operates as it does today. In a recent conversation, Veatch expressed her desire to document the origins of generative AI to provide insight into the intense cycle of industry hype we currently experience. However, she first had to navigate the intentional confusion surrounding the term “artificial intelligence.”

“To use the term ‘artificial intelligence,’ we need to understand what it actually means,” Veatch stated during our video call. “The reality is, it doesn’t mean anything; it’s a marketing term that has always been misleading. It’s crucial to be clear about the language we use and its meanings.”

*Ghost in the Machine* emphasizes that the term “artificial intelligence” was first introduced in 1956 by computer scientist John McCarthy, who sought funding for his projects. The documentary positions this coinage as just one significant moment in a timeline that traces back to Victorian England and the origins of eugenics. Francis Galton, a cousin of Charles Darwin and the founder of eugenics, promoted the discredited notion that humanity could be improved through the elimination of “inferior” races.

While Galton made some valuable contributions to academia, Veatch highlighted the importance of recognizing how his white supremacist beliefs influenced the social sciences of his time. Although Galton and his protégé Karl Pearson were not directly involved in early computational machines, Galton’s work on multidimensional modeling—used to assess the attractiveness of women from different races—shaped Pearson’s development of statistical tools like logistic regression, which is fundamental to modern machine learning.

Galton and Pearson helped normalize the idea that different races could be quantitatively measured, leading to the belief that human intelligence could be assessed similarly, reinforcing the notion that human brains function like machines. Veatch noted that this perspective significantly contributed to the public’s acceptance of the fantastical concept of artificial intelligence.

“What surprised me during my research was how quickly the topic of superintelligence leads you to the problematic history of race science, which is deeply embedded in this technology,” Veatch explained, emphasizing that these ideas are “soaked” in eugenic thought.

Instead of attempting to refute the notion that generative AI models produce harmful ideologies due to their training data—a concept known as “GIGO” (garbage in, garbage out)—*Ghost in the Machine* uses historical analysis to shed light on why the companies behind this technology seem uninterested in addressing its current issues. This context helped Veatch make sense of her own unsettling experiences with generative AI, particularly when she was experimenting with an early version of Sora in an artist community on Slack. She recalls the group being friendly until a woman of color raised concerns about how the model consistently whitewashed her when generating images based on her photos.

“It kept her braids and fashion, but when she prompted it to create an art gallery scene, the program interpreted that as a ‘white space,’” Veatch recounted. “My reaction was, ‘what the hell,’ and I tried to explain to the group that this was a significant issue with the software.” However, her post went unanswered. “Normally, there are tons of reactions in that Slack, but this time, there was silence.”

Veatch decided to reach out to OpenAI directly to report the “racist, sexist, and misogynistic” outputs she observed, including bizarre transformations of women in generated scenes. She expected the company to treat her feedback as a critical issue to address before promoting Sora further, but instead, her concerns were dismissed.

“The response I received was essentially, ‘This is cringe to bring up; there’s nothing we can do to change it,’” Veatch recalled.

This experience ignited Veatch’s determination to understand why generative intelligence consistently exhibits such troubling behavior. Initially skeptical that discussions with white paper authors could lead to a compelling documentary, her perspective shifted as she recognized a clear connection from Galton’s eugenic statistics to contemporary generative AI companies.

The voices featured in *Ghost in the Machine*—a mix of AI researchers, historians, and critical theorists—argue that nearly every aspect of the AI landscape has been profoundly shaped by its historical ties to scientific fields that supported discriminatory ideologies. When asked if she had considered interviewing the leaders of the companies critiqued in her film, Veatch laughed, noting that gaining such access would require her to navigate ideological compromises that would undermine her film’s integrity.

“There’s this notion that these people won’t trust just anyone,” Veatch said. “Of course, I wouldn’t expect them to trust me. I don’t want them in the film, and they already have plenty of media exposure. Am I supposed to hug Sam Altman on camera? That wouldn’t be an honest portrayal of this technology; it would be propaganda.”

*Ghost in the Machine* will be available for streaming from March 26th to March 28th before airing on PBS later this fall.