Ghost in the Machine Valerie Fitch Don’t Drink the AI ​​Kool-Aid

Like many people, director Valerie Fitch was fascinated when OpenAI debuted its generative AI model Sora for text-to-video to the public in 2024. Although she didn’t fully understand the technology, she was interested in what it could do, and saw that other artists were building online communities to share their new AI creations. The hope of connecting with people drew Fitch to the field of artificial intelligence, but once she got there, she was shocked to see how often the technology produced images laced with racism and sexism.

Fitch was most disturbed by the way her new AI-enthusiastic peers seemed not to care that the machine they rallied around spewed bigoted garbage and hate without being explicitly asked to do so. The strange situation pushed Fitch away from her early experience with artificial general intelligence. But it also inspired her to do so Ghost in the machinea new documentary about the technologies and schools of thought that laid the foundation for the existence of artificial intelligence.

Instead of focusing on the potential (albeit highly unlikely) benefits to society that the generation of AI accelerators swear is just around the corner, Ghost in the machine Explores the history of technology to explain why it worked the way it does now. When I recently spoke with Fitch about the film, she told me that she wanted to chronicle the rise of artificial intelligence to give people a clear view of the extremely intense cycle of industrial hype we’re currently experiencing. However, it first had to overcome AI companies’ deliberate obfuscation of the entire concept.

“In order to use the phrase ‘artificial intelligence,’ we have to know what that phrase means,” Fitch told me over a video call. “The truth is it doesn’t mean anything; it’s a marketing term and it always has been. It’s a completely misleading and stupid phrase that has taken on its own cultural meaning, and I think being really clear about the words we use and the meaning of those words is essential.”

like Ghost in the machine He repeatedly asserts that “artificial intelligence” was originally coined in 1956 by computer scientist John McCarthy when he was trying to secure more funding for his projects. But the documentary presents the coining of the term as one of many important points in a timeline that actually begins in Victorian England with the birth of eugenics. In addition to being Charles Darwin’s cousin, Francis Galton was the originator of eugenics — the discredited racist belief that humanity could be improved by systematically eliminating “inferior” (read: non-white) races.

While Galton certainly provided some His useful contributions to academia Fitch explained in our interview that it is important not to downplay the fact that it was his deep-rooted white supremacist beliefs that influenced the social sciences of that era. Galton and fellow eugenicist/student Karl Pearson were not directly involved in the development of early arithmetical machines. But Galton’s foundational work in multidimensional modeling—a technique he used while measuring the attractiveness of African and European women—helped shape Pearson’s thinking as he developed statistical tools like logit regression, a core component of modern machine learning.

Galton Pearson Help normalize the idea That people of different races were fundamentally different in quantifiable ways. This kind of racist thinking is what led Galton and his peers to believe that human intelligence can be measured, and that human minds work much like machines. Fitch says this leap played a major role in convincing the public of the fantastical idea of ​​artificial intelligence.

“What was really surprising to me during my initial delve into all of this is that when you look at the question of superintelligence as a documentarian or journalist, it doesn’t take long before you’re banging your forehead against a low door frame for race science, because it’s built into this technology,” Fitch said, explaining that these concepts are “steeped” in eugenics thinking.

Instead of trying to refute the idea that general AI models produce hateful ideology because they have been trained in it (a concept known as “GIGO” – garbage in, garbage out), Ghost in the machine He uses his historical analysis to explain why companies making this technology are not interested in addressing their current problems. This historical context helped Fitch make sense of some of her troubling experiences with the AI ​​gen, when she was playing with an early version of Sora in Slack for artists. Fitch recalls that the group was a friendly, welcoming place until another member — a woman of color — began voicing concerns about how the model whitewashed her every time she asked her to create images based on photos of herself.

“She kept her braids and kept her fashion, but she was pushing herself into an art gallery, which the show understood as ‘white space,'” Fitch explained. “I was like, ‘What the heck,’ and I tried to explain to the group how this was an issue with the show itself.” No one else in the group reacted to her post. “This was like Slack, where there are always dozens of screaming koala emoji reactions in each post. But this time, there was nothing.”

Veatch took it upon herself to contact OpenAI directly to alert the company about “how much racism, sexism, and misogyny it was seeing — output where women start growing extra breasts and working out after a couple of rounds of creating a scene.” Fitch believed that OpenAI would consider this a serious bug worth fixing before encouraging more people to adopt Sora into their lives; Instead, the company ignored her concerns.

“The feedback I got was basically, ‘It’s very annoying to bring this up; “There is nothing we can do to change it.”

This situation ignited a passion within Fitch to find out why so many different forms of generative intelligence continually behaved in such ugly and disturbing ways. At first, she didn’t really think that doing Zoom calls with authors of papers on the technology could turn into a compelling documentary, but that changed when she started to see a clear line from Galton’s eugenics statistics work to the clothes of the modern generation of artificial intelligence.

The sounds that appeared in Ghost in the machine — a mix of AI researchers, historians, and critical theorists — make a compelling case that every aspect of the AI ​​field has been profoundly influenced by its historical connections to areas of science that were designed to uphold discriminatory worldviews. When I asked Fitch if she would be interested in speaking directly with company presidents Ghost in the machine I laughed. Gaining that kind of access, she said, would require her to go through all kinds of ideological exercises and make compromises that would make her film complicit in the harms of the AI ​​general.

“There’s this idea that, you know, these people aren’t going to trust anyone,” Fitch said. “Yes, nothing, and I certainly hope they don’t trust me. I don’t want them in the movie talking enough to the media. Am I going to hug Sam Altman on camera? Is this an honest film about this technology? This is propaganda.”

Ghost in the machine will be available to Stream via Kinemfrom March 26 to March 28 before airing on PBS sometime this fall.