Self-funded documentary Ghost in the Machine takes aim at AI’s roots in eugenics

Valerie Veatch, director of Ghost in the Machine, a documentary that looks past the AI hype, says the technology is rooted in eugenics, which still matters today, because the act of creating a data point is deeply political.

The movie premiered at the Sundance Film Festival on January 26th, 2026, and is now available on PBS Documentaries’ YouTube Channel.

Veatch isn’t a novice at exploring technology’s impacts on our lives. Her 2014 documentary Love Child investigates internet addiction, while Me at the Zoo, released in 2012, follows the life of video blogger Chris Crocker.

She tells Cybernews that the idea of creating a documentary about AI was born after a friend signed her up for an artists’ working group with OpenAI’s Sora, a now-retired tool for generating videos.

After she found Sora’s outputs highly disappointing, Veatch started learning more about how LLMs work, which led to revelations about how the technology is being built around extraction, resources, colonialism, and control.

AI’s 20th-century racist roots still matter

The documentary explores how statistics and machine learning – a subset of AI – stems from eugenics, a discredited ideology of racial improvement and planned breeding, whose followers believed in the genetic superiority of white Europeans.

Perceived biological and intellectual differences between races were used to justify the British Empire's control of foreign lands and people, as well as the extraction of their resources.

The documentary takes a deeper look at Charles Spearman, an English scientist and father of modern statistics, who developed many statistical tools used today to defend and promote eugenics.

Spearman theorized about the g factor, short for general intelligence factor, which allowed for the measurement and ranking of intelligence, a notion that later laid the groundwork for discriminatory laws in the US, which were later adopted by the Nazis in Germany.

While eugenics has long disappeared from mainstream science, Ghost in the Machine explores how the obsession with measuring intelligence and eugenicist ideas still echo in the field of AI.

Abeba Birhane, who is in the movie, says in her work that statistics may be mathematically neutral, but the act of creating a data point or the act of statistics is deeply political. Statistics have always been racial politics.

Valerie Veatch, a director of Ghost in the Machine

Using the word “intelligence” in the term AI, computer scientists look to intelligence research, which has been funded by fringe figures, resulting in problematic notions that still linger today.

The AI industry’s rhetoric increasingly focuses on different levels of intelligence. The theoretical general artificial intelligence (AGI), for example, would match human cognitive abilities across tasks, while superintelligence (ASI) would exceed human intellectual performance.

President Donald Trump recently announced the renaming of AI to “super intelligence” in official US documents. However, the POTUS was more concerned about the word “artificial” making AI sound fake than about changes in the technology’s capabilities.

Veatch points out that LLMs have been shown to produce ableist, sexist, and racist outputs. When AI reduces things to similar data points, and the next statistically most likely syllable in a word or pixel in a picture, she says, this all is further intensified.

“If we’re using these tools for vibe writing or finding ideas, we are allowing this highly racialized point of view into our work,” she says.

Debunking the AI doomer narrative

Veatch tells Cybernews that the movie was self-funded, resulting in “a sort of punk energy,” as she puts it.

“I just grabbed my web camera and started interviewing people, and I let go of the need for it to be aesthetic,” she says.

The movie combines archival footage, excerpts from news reports and podcasts, and interviews with nearly 40 experts, including Professor Emily M. Bender, who co-invented the term “stochastic parrots” to describe LLMs that lack understanding of their outputs.

Veatch says not including interviews with leaders of AI companies was a deliberate decision. Partly because she didn’t think they would have anything new to say, but also because access to them would have required an unacceptable level of complicity.

Meanwhile, executives at leading AI labs, OpenAI and Anthropic, have recently called for slowing the development of the technology, following a string of cybersecurity incidents and growing warnings from industry insiders about the potential for AI to wipe out humanity.

Veatch is skeptical of these claims, noting that OpenAI initially withheld the release of GPT-2 in 2019, citing concerns that it might be too dangerous.

Industry insiders’ warnings make Ghost in the Machine even more timely, as experts interviewed for the documentary reject these inflated fears and consider terms such as AGI and ASI to be a new package for selling an autonomous thinking machine that may never come.

The documentary also takes a closer look at how the AI industry exploits people in the Global South who work on data annotation, only to be left with low pay and psychological trauma resulting from watching graphic materials.

Apocalyptic future is not inevitable

Most of the footage in the movie carries a “non-AI” label, which Veatch says was used as a punchline during editing.

However, it may also serve to question AI narratives, as some of the things tech company leaders say at the end of the film “become increasingly untethered from reality,” according to Veatch.

The movie invites the normalization of AI refusal and the dehumanization that comes with big tech efforts to sell the technology, hidden behind the marketing term "AI."

Veatch says, “We don't have to participate in this notion of an inevitable future that's going to sweep us away.”