Challenging AI hype narratives with director Valerie Veatch

Setting out on the warpath

In October 2024, Veatch became involved in an early access programme for artists organised by OpenAI, which would see creatives interact for the first time with the firm’s text-to-video generative AI (GenAI) model, Sora.

The rhetoric surrounding the event was immediately striking: AI would liberate the filmmaker and democratise the production process.

Veatch, however, left shocked: “There should be a special word for the horrible thing that happens when something comes up that you didn’t prompt. Immediately, I noticed horrendous images that were hugely sexist and hugely racist.”

Although Veatch raised concerns directly with OpenAI representatives, she left with the impression that “it’s not something they really want to address”.

Left with “the ick”, Veatch set out on the warpath, speaking with dozens of critical AI experts over a nine-month period. At first, there was no intention of creating a film, but as the conversations piled up, Veatch noticed a cohesive story starting to emerge.

“Everybody’s telling parts of the same story, like different notes of the same song,” she says. Eventually, Veatch ended up with around eight hours of select footage, which she then began condensing into a feature-length documentary.

“I felt this film could be a vessel for a lot of the existing work in the tech critical space. All the ingredients inside the machine learning stew come from the framework of colonialism, extraction, control, so maybe it could centre some of these ideas.”

However, despite plugging into a rich body of existing critiques on AI, pitching the film was no easy task: every production and distribution company was simply too enthused about the technology, and refused to back a project that aimed to question its foundational pillars.

“Without exception, every single person I approached told me, ‘Our company is launching an AI initiative, and we’re really excited about that’,” she says.

“Throughout my career, I’ve done stuff that is tech critical. There’s always that flavour, but I was really surprised at the mainstream mentality, at how you couldn’t really put something out there that said ‘AI sucks’.”

In the end, the self-funded project is being distributed via a combination of film festivals, virtual cinema screenings and television broadcasts, and is opening in UK theatres on 5 June 2026.

Breaking the hype: AI as it actually exists

For Veatch, any project that engages in the hype rhetoric of the booster and doomer misses the realities of the technology as it exists today.  

From an environmental standpoint, it means missing how the technology is “actively destroying the planet and displacing water resources”, while from a labour perspective, it means ignoring the human toll of delivering AI.

Highlighting, for example, how Waymo recently admitted to Congress that its autonomous vehicles are operated remotely by workers in the Philippines who are getting paid a dollar a day, or how data labellers in Kenya are suffering mental health repercussions as a result of their gig work for AI firms, Veatch says there is a “level of labour obscurity” that is staggering.

“The massive amount of money that is required for these systems to operate certainly doesn’t go to those folks,” she says.

Moving away from the production side of things, Veatch adds that, in its real-world operation, AI also leads to an “algorithmic intensification” of the existing negative patterns within society.

“Going into the structure of a dataset, it certainly does not represent anything other than a very specific world view that sources that dataset, so it’s funny when people turn around and imagine that these systems would produce anything different,” she says.

“If you [use algorithms or AI systems] to send police to a neighbourhood, they’re going to find more crime, that neighbourhood’s going to read as having a higher crime index, and more police will be sent. It’s a self-fulfilling thing.”

Ultimately, Veatch concludes that when we focus on doomer-hype narratives, “it obscures the human rights that are displaced through the environmental and labour components of the technology … and we miss the real harms of the algorithmic system”.

Planetary technicity/global dehumanisation

For Veatch, it is no coincidence that AI – an extractivist technology that relies on the exploitation of hidden human labour and the unprecedented accumulation of data and natural resources – is taking off during a period of mass global dehumanisation.

From Sudan and Congo to Nagorno-Karabakh and Gaza, genocidal violence has reached a scale that most people alive today will not have seen in their lifetimes. And this is the same time as AI is being revered as a messianic second coming.

“We have this mass scale dehumanisation, and at the same time have this elaborately funded, hugely propagandistic project to hype machines that can think and hype this idea of super intelligence,” she says.

“It’s interesting on a macro level how these trends are firing. It’s much easier to imagine AI as this ‘God thing’ than to imagine an end to children starving and dying.”

She adds that the computational, algorithmic logic of AI also represents a cultural tendency towards dehumanisation, as it reduces the complex lives of human beings to numbers or data points.

It is no mistake then that AI is applied by governments to areas where the people on the receiving end of it are deemed as more expendable than others: immigration, welfare systems, policing and the military.

The advent of AI, therefore, creates a utilitarian feedback loop, where human thinking becomes increasingly cold and mechanical, and machine thinking can be increasingly relied on because it has become so entwined with how people view one another as distant, atomised others.  

In applying AI to these situations, it then becomes easier to deny someone their benefits, subject them to increasingly intense surveillance, or even kill them.

“One thing that these systems are good at is causing harm,” she says, linking back to the idea of algorithmic intensification: “If you plug an LLM [large language model] into a targeting system, which is scraping a population for who’s the likely target for your next drone attack, as we see playing out like every day in the Middle East, what do you think will happen?

“These systems are not gaining consciousness, they’re not going to cure cancer, they’re not going to create any new ideas. But what they are going to do is sort data in increasingly harmful ways.”

A question of power: Articulating a culture and politics of refusal

Speaking in 2010, free market libertarian and PayPal billionaire Peter Thiel vocalised a dream of the future that has come to encapsulate the current moment with AI.

“The initial founding vision [of PayPal] was that we were going to use technology to change the whole world, and basically overturn the monetary system of the world,” he said at the time. “We could never win an election on getting certain things because we were in such a small minority, but maybe you could unilaterally change the world without having to constantly convince people and beg people and plead with people who are never going to agree with you through technological means.”

I hope we can build architectures and frameworks for refusing AI, and promote the cultural permission to refuse AI. Culturally, we can reject AI and just flat-out stop using it

The idea promoted by Thiel – who is one of the richest 100 people on the planet – that technology offers an “incredible alternative to politics”, is not something that should be read as an idiosyncrasy, but something that characterises an increasingly widespread, technocratic approach to politics across the world.

For Veatch, it is telling that AI evangelists throughout Silicon Valley consciously view the technology as a form of power: “Fascistic ways of ordering and acting are brought about through infrastructure, more than they are through symbolic means.”

AI is therefore not simply a technology, but a form of techno-politics, whereby technical systems and political practices co-constitute and reinforce one another in a feedback loop that creates distinct forms of technocratic power and governance.

In essence, AI is a highly concentrated, hierarchical technology, because the society deploying it is defined by substantial power concentrations, which in turn leads to even greater centralisation and a deeper embedding of those existing hierarchies that wield the capital, infrastructure and data to run it in the first place.

While there is the latent possibility of decentralised control of technical infrastructure, the current technopolitical reality of modern states under capitalism is prefigured by institutions (whether private or public) that are top-down and hierarchical.

In highlighting how AI exacerbates and entrenches existing problems in society, Veatch says refusal is a politically underexplored avenue for dissent, and that there is power in just saying no.

“I hope we can build architectures and frameworks for refusing AI, and promote the cultural permission to refuse AI,” she says. “Culturally, we can reject AI, and we can just flat-out stop using it.”

Highlighting how many prominent critics or non-governmental organisations caveat their criticisms of technology with phrases like, “Well, I’m not a Luddite/technophobe, but…”, Veatch says the idea of outright AI refusal is currently difficult for many to swallow.

The positive is that “culture is quick, culture is ours, and we can shape it”.