Similar to numerous individuals, filmmaker Valerie Veatch found herself fascinated when OpenAI initially unveiled its Sora text-to-video generative AI model to the public in 2024. Despite not fully grasping the underlying technology, she harbored a keen curiosity about its potential. She observed other creators establishing virtual collectives to showcase their fresh AI-generated works. The prospect of connecting with fellow creators drew Veatch into the AI domain, yet once immersed, she was appalled by the regularity with which the technology would render depictions rife with racism and sexism.
What further perturbed Veatch was the apparent indifference of her fellow AI aficionados, who seemed unconcerned that the system they championed disgorged offensive, bigoted material even without explicit instruction. This peculiar circumstance caused Veatch to withdraw from her initial forays into generative AI. However, it simultaneously spurred her to create Ghost in the Machine, a new documentary delving into the methods and intellectual currents that formed the foundation of gen AI’s genesis.
Rather than concentrating on the hypothetical (albeit improbable) boons for humanity that proponents of rapid AI advancement claim are on the horizon, Ghost in the Machine delves into the technology’s past to illuminate its current operational mechanisms. When I recently conversed with Veatch concerning the documentary, she expressed her aim to document gen AI’s origins, thereby offering individuals an unobstructed insight into the exceptionally vigorous period of industrial enthusiasm we’re presently experiencing. Initially, nonetheless, she needed to penetrate the deliberate concealment of the entire concept by AI corporations.
“To employ the term ‘artificial intelligence,’ we must genuinely understand its significance,” Veatch conveyed to me during a video call. “In reality, it conveys no substantive meaning; it functions as a promotional label, and always has. It’s an utterly deceptive, ill-conceived expression that has acquired its own distinct cultural interpretation, and I believe that achieving genuine clarity regarding the words we use and their meanings is paramount.”
As Ghost in the Machine consistently emphasizes, the phrase “artificial intelligence” was first formulated in 1956 by computer scientist John McCarthy while he endeavored to obtain greater financial backing for his undertakings. Yet, the documentary portrays the term’s creation as merely one significant juncture among several on a chronology that truly commences in Victorian-era England with the dawn of eugenics. Beyond his kinship with Charles Darwin, Francis Galton was the progenitor of eugenics — the prejudiced and disproven conviction that humankind could be enhanced via the methodical eradication of “substandard” (specifically, non-Caucasian) ethnicities.
Though Galton indeed offered certain valuable contributions to academia, during our discussion, Veatch clarified the imperative of not downplaying how his entrenched white supremacist convictions influenced the social sciences of that period. Galton and his colleague in eugenics and protégé, Karl Pearson, lacked direct participation in the creation of nascent computing devices. Nevertheless, Galton’s pivotal efforts in multidimensional modeling — a method he employed for assessing the appeal of African and European females — influenced Pearson’s conceptualization while he devised statistical instruments such as logistic regression, which remains a core constituent of contemporary machine learning.
“Should I embrace Sam Altman for the cameras? Would that constitute an honest portrayal of this technology? Such an act would be mere propaganda.”
Galton and Pearson contributed to the mainstream acceptance of the notion that individuals from diverse ethnic backgrounds possessed inherent, measurable distinctions. Such prejudiced ideology prompted Galton and his contemporaries to conclude that intellectual capacity was quantifiable and that human minds operate akin to mechanical systems. According to Veatch, that conceptual leap was instrumental in persuading the populace about the fanciful concept of artificial intelligence.
“What was particularly astonishing to me upon my initial immersion in this subject matter was how, when one, as a documentarian or journalist, examines the issue of superintelligence, one quickly confronts the entrenched presence of race science, as it’s intrinsically embedded within this technology,” Veatch stated, clarifying that these notions are thoroughly imbued with eugenic ideology.
Instead of attempting to refute the notion that generative AI models yield prejudiced content due to their training data (an idea popularly termed “GIGO”—garbage in, garbage out), Ghost in the Machine employs its historical examination to elucidate why corporations developing this technology appear so unconcerned with tackling its contemporary problems. Such historical background enabled Veatch to comprehend certain disquieting encounters she had with generative AI, from when she experimented with an initial iteration of Sora within a creatives’ Slack channel. Veatch recalled the collective as a congenial, hospitable environment, right up until another participant — a woman of color — started expressing apprehensions regarding how the model de-racialized her whenever she prompted it to generate visuals using her own photographs.
“It retained her braided hairstyle and preserved her attire, yet she was imagining herself within an art exhibition, an environment the software interpreted as a ‘Caucasian domain,’” Veatch elaborated. “My immediate response was one of profound shock, and I attempted to clarify to the collective that this constituted a genuine flaw within the software itself.” No other member of the group interacted with her submission. “This was a Slack where, typically, scores of expressive koala emoji responses would accompany each message. However, on this occasion, utter silence prevailed.”
Image: Independent Lens
Veatch decided to personally reach out to OpenAI, aiming to inform the company about the “racist, sexist, and misogynistic outputs [she] was encountering — generations where women would start growing extra tits and twerking after merely a couple of scene generations.” Veatch anticipated that OpenAI would perceive this as a significant flaw deserving of correction before encouraging wider adoption of Sora; nevertheless, the firm disregarded her apprehensions.
“The response I received was essentially, ‘This is quite awkward to mention; we are incapable of altering it,’” Veatch recounted.
That incident ignited a strong resolve in Veatch to investigate why numerous varieties of generative intelligence invariably operate in such undesirable, problematic manners. Initially, she hadn’t truly considered that conducting video conferences with the creators of academic papers concerning the technology might be transformed into a captivating documentary. However, this evolved as she began to discern a direct connection between Galton’s eugenic statistics work and contemporary generative AI enterprises.
The perspectives included in Ghost in the Machine — a combination of AI researchers, historians, and critical theorists — present a persuasive argument that virtually every aspect of the AI domain has been deeply impacted by its historical ties to scientific disciplines developed to uphold biased worldviews. When I inquired if Veatch had ever been keen on conversing personally with the leaders of the corporations Ghost in the Machine challenges, a chuckle escaped her. Obtaining such entry, she stated, would necessitate her undergoing numerous ideological contortions and entering into concessions that would render her documentary an accomplice to the detriments of generative AI.
“There’s this notion, you see, that these individuals will not repose faith in just anyone,” Veatch remarked. “Naturally, and I genuinely wish they would not place their confidence in me. I do not desire their presence in the documentary, and they already communicate extensively with the press. Shall I embrace Sam Altman for the camera? Would that constitute an authentic depiction of this innovation? That is pure indoctrination.”
Ghost in the Machine can be accessed for streaming through Kinema between March 26th and March 28th, prior to its broadcast on PBS later this autumn.
{content}
Source: {feed_title}