Ghost in the Machine Exposes the Dark Eugenics History of AI at Sundance

Valerie Veatch’s documentary Ghost in the Machine premiered at the 2026 Sundance Film Festival on January 26, 2026, in Park City, Utah, reframing artificial intelligence not as a neutral software evolution, but as a direct descendant of eugenics, statistical racism, and industrial domination. Described by critics as a gripping 110-minute essay film screened in the festival’s NEXT section, the project investigates how historical figures like Karl Pearson and William Shockley laid the mathematical and ideological foundation for modern machine learning technologies. Featuring testimony from more than 30 critical researchers—including Emily M. Bender, Adam Becker, and Tiera Tanksley—the film exposes how contemporary tech accelerationism remains tethered to nineteenth-century ideas of human quantification and control.

Tracing the Eugenics Roots of Modern Machine Learning

Silicon Valley often operates under the assumption that it exists in an ahistorical vacuum, according to computer scientist Adam Becker, who spoke at the post-premiere Q&A at The Yarrow Theatre. Yet, as Ghost in the Machine demonstrates through archival material and expert analysis, the core statistical models powering today’s large language models trace directly back to the work of Karl Pearson. In the documentary, researcher Ezekiel Dixon-Román explains that these early mathematical frameworks emerged explicitly out of eugenics. The film links those historical practices to William Shockley—known as the man who brought silicon to Silicon Valley—highlighting televised appearances where he proposed voluntary sterilization bonus plans.

The Marketing Myth of General Artificial Intelligence

Beyond its historical excavations, Veatch’s documentary argues that artificial intelligence functions primarily as a marketing tagline rather than a realized technological breakthrough. Computer scientist John McCarthy is featured in archival footage explaining how he and his contemporaries coined the term in the mid-twentieth century simply as a simplified shorthand to secure funding from potential bankrollers. Veatch herself was initially drawn into the subject when an AI company invited her to test a new filmmaking tool touted as a total democratization of the medium. When she began questioning the systemic biases baked into the code, representatives told her not to ask those questions. That corporate stonewalling pushed her to take those inquiries directly to independent researchers.

Visual Deception and the Environmental Toll of Tech Accelerationism

The documentary incorporates generated imagery to illustrate the uncanny outputs of modern machine learning tools, transforming archival motifs into distorted reflections of themselves. This visual mimicry serves as a pointed critique of the massive resource extraction required to sustain the industry. Unopened water bottles sitting between tech executives at panels morph into digital mutants on screen, symbolizing the vast amounts of freshwater required to cool supercomputing centers. To help audiences parse reality from synthesis during her production process, Veatch adopted a low-tech countermeasure, distributing physical “Not AI” pins to attendees following the Sundance screening.

Mounting an independent documentary circuit or managing high-profile festival premieres of this scale involves massive logistical coordination.

Resistance, Regulation, and the Human Element

Despite the grim picture painted by the accumulation of wealth and power among tech elites, the film concludes on a cautiously optimistic note. In the documentary, Tiera Tanksley pushes back against the myth that the future is predetermined, while Bender advocates for rigorous labor protections, stronger regulatory oversight, and a return to community-driven resistance. By questioning the necessity of automating specific human responsibilities, the film suggests that reclaiming the human component remains the most effective form of pushback against technological determinism.