Aura Kasih
A new documentary, "Ghost in the Machine," directed by Valerie Veatch, probes the historical underpinnings of generative artificial intelligence, arguing that its current manifestations of bias and problematic outputs are not accidental but deeply rooted in discredited scientific theories, particularly eugenics. The film challenges the prevailing narrative of AI as a purely technological advancement, instead positing that its foundational concepts are intertwined with early 20th-century attempts to quantify and categorize human populations based on race and perceived intelligence.
Veatch’s initial engagement with generative AI, specifically OpenAI’s Sora text-to-video model, was driven by the typical curiosity that accompanied the technology’s public debut in 2024. Like many artists and creators, she was drawn to the potential for new forms of expression and the burgeoning online communities forming around AI-generated content. However, her exploration quickly soured as she encountered a persistent and disturbing tendency for the AI models to produce outputs saturated with racist and sexist imagery, often without explicit prompting. This experience was compounded by what Veatch perceived as a troubling apathy among fellow AI enthusiasts, who seemed unwilling to confront the problematic nature of the technology they championed. This disconnect between the promise of AI and its often discriminatory realities became the impetus for "Ghost in the Machine," a film that endeavors to trace the intellectual lineage of generative AI, examining the historical currents that shaped its development and continue to influence its behavior.
The documentary distinguishes itself from the typical discourse surrounding AI, which often focuses on its potential societal benefits or the speculative future of superintelligence. Instead, Veatch’s work delves into the historical context, seeking to illuminate why current generative AI systems exhibit the biases they do. In an interview, Veatch articulated her intention to demystify the genesis of generative AI, providing a clearer perspective on the intense cycle of industry hype that currently surrounds it. A crucial element of this demystification, she emphasized, involves dissecting the very language used to describe these technologies, particularly the term "artificial intelligence."
Veatch contends that the phrase "artificial intelligence" is more a marketing construct than a precise scientific descriptor. Coined in 1956 by John McCarthy in an effort to secure funding for his research, the term, according to Veatch, has acquired a cultural meaning divorced from its literal implications. She stresses the importance of linguistic clarity, asserting that the ambiguity surrounding "artificial intelligence" obscures a more complex and problematic historical trajectory.
"Ghost in the Machine" meticulously charts this trajectory, beginning not with the advent of computing but in Victorian-era England with the emergence of eugenics. The film identifies Francis Galton, Charles Darwin’s cousin and a proponent of eugenics, as a pivotal figure. Eugenics, a scientifically discredited and morally reprehensible ideology, posited that humanity could be improved through the systematic elimination of so-called "inferior" races. While acknowledging Galton’s legitimate contributions to fields like statistics, Veatch underscores the pervasive influence of his white supremacist beliefs on the social sciences of his era.
Although Galton and his protégé, Karl Pearson, were not directly involved in the development of early computational machines, Galton’s foundational work in multidimensional modeling had a profound impact. His methodologies, employed in attempts to quantify human physical characteristics, influenced Pearson’s development of statistical tools such as logistic regression. This statistical technique is a fundamental building block of modern machine learning algorithms, forming a crucial link between eugenicist thought and contemporary AI.
The film argues that Galton and Pearson were instrumental in normalizing the idea that racial groups exhibit quantifiable, fundamental differences. This racist framework underpinned the belief that human intelligence could be measured and that the human brain operated in a manner analogous to a machine. This conceptual leap, Veatch posits, was essential in popularizing the ambitious, and arguably fantastical, notion of artificial intelligence.
"What was really surprising to me during my initial dive into all of this was how, when you look at the question of superintelligence as a documentarian or journalist, it doesn’t take long before you smack your forehead into the low doorframe of race science, because it’s baked into this technology," Veatch stated, highlighting how these concepts are "soaked" in eugenic thinking.
Instead of merely attributing the generation of hateful ideology by AI models to flawed training data (the "garbage in, garbage out" principle), "Ghost in the Machine" utilizes its historical analysis to explain the apparent disinterest of AI companies in addressing these present-day issues. This historical context provides a framework for understanding Veatch’s own disquieting experiences with early versions of generative AI. During her time experimenting with Sora within an artists’ Slack community, Veatch observed a stark shift in group dynamics. What began as a welcoming environment soured when a member, a woman of color, raised concerns about the AI consistently whitewashing her likeness when prompted to generate images based on her own photographs.
"It kept her braids and it kept her fashion, but she was prompting herself into an art gallery, which the program understood to be a ‘white space,’" Veatch explained. "My reaction was ‘what the fuck,’ and I tried explaining to the group how this was really a problem with the software itself." Her attempt to initiate a discussion was met with silence, a notable contrast to the typically vibrant and reactive nature of the online forum.
Veatch’s direct outreach to OpenAI, informing them of the "racist, sexist, and misogynistic outputs" she was witnessing, yielded a dismissive response. The company, she recounted, characterized her concerns as "cringe" and stated there was "nothing we can do to change it," effectively prioritizing product development and market adoption over addressing fundamental issues of bias.
This interaction galvanized Veatch’s resolve to investigate the persistent and pervasive problematic outputs of generative AI. While initially uncertain if discussions with white paper authors could form the basis of a compelling documentary, her perspective shifted as she identified a clear intellectual lineage connecting Galton’s statistical work in eugenics to contemporary generative AI firms.
The documentary features a diverse array of voices, including AI researchers, historians, and critical theorists, who collectively argue that nearly every aspect of the AI landscape has been profoundly shaped by its historical connections to scientific fields that once served to uphold discriminatory social structures. When questioned about her interest in directly interviewing the leaders of companies criticized in the film, Veatch responded with amusement. She indicated that securing such access would necessitate ideological compromises that would undermine the film’s integrity and potentially render it complicit in the harms perpetuated by generative AI.
"There’s the idea, you know, these people won’t trust just anyone," Veatch stated, emphasizing her deliberate distance from the industry. "Yeah, no shit, and I certainly hope they wouldn’t trust me. I don’t want them in the film and they already speak enough to the media. Am I going to hug Sam Altman on camera? Is that a truthful film about this technology? That’s propaganda."
"Ghost in the Machine" aims to provide a critical lens through which to view the rapid advancements in generative AI. By foregrounding the historical context and the intellectual roots of the technology, the film challenges the uncritical embrace of AI and calls for a more rigorous examination of its ethical implications. The documentary suggests that the allure of generative AI, often framed as a purely technological leap forward, is inextricably bound to a problematic past, demanding a more nuanced and historically informed approach to its development and deployment. The film is scheduled for streaming via Kinema from March 26th to March 28th, with a subsequent broadcast on PBS in the fall.
Ridwan Kamil
Lisa Mariana
