is a reporter focusing on film, TV, and pop culture. Before The Verge, he wrote about comic books, labor, race, and more at io9 and Gizmodo for almost five years.
There has been a recent deluge of generative AI videos featuring uncannily fresh-faced teens waxing nostalgic about how much better the world was during the ’80s and ’90s. As the AI youths smize and show off their period-specific haircuts, the clips cut to dreamlike footage of sun-drenched cul-de-sacs and vintage cars while songs like “Everybody Wants to Rule the World” and tracks inspired by the Donkey Kong Country soundtrack play in the background. It’s all very weird — like bragging that you peaked in high school.
As strange as the videos are, there is a relatively easy to understand logic at work here. On one level, this content appeals to people’s fascination with the past — especially younger viewers whose lack of firsthand experience with these eras can make it easier to overlook the anachronistic details generative AI models are prone to including in their video output. But these videos are also conjuring an idealized vision of the past where everyone is beautiful, most people are white, and they all have inexplicable knowledge about how stressful life in 2025 is going to be. This kind of nostalgia is a neocon fantasy for people allergic to cracking open history books.
For some reason, Fred Rogers is often the focus of these clips where you can see him rapping with Tupac, perving on women like Marilyn Monroe, and showing off a closet full of guns. None of these deepfakes are especially convincing and most of them still have watermarks indicating that they were created with OpenAI’s Sora model. But as terrible as this slop is, it’s everywhere, and the view counts suggest that — regardless of whether it’s out of love or hate or ambivalence — people can’t stop themselves from watching. At least, that’s probably what the team behind OpenAi’s recently launched social video app wants you to think.
It’s fairly obvious what OpenAI stands to gain from flooding the internet with Sora-generated videos. The content is another way for the company to promote its technology and normalize the idea of people clocking in at the slop factory as a way of entertaining themselves. That seems to be the endgame for the Sora app, where generating a video is as simple as typing a few sentences into a prompt box. OpenAI and its competitors all want to be perceived as wellsprings from which a new, revolutionary kind of art has emerged — one that gives people the ability to express their creativity in ways that were not possible before.
The people making these videos like Jake Paul, Snoop Dogg, and Shaquille O’Neal have clearly bought into that idea, or at least been paid to pretend they have in order to convince their gullible fans that mainlining slop from a trough is cool, actually. But when you watch enough of this stuff (which isn’t a lot), what becomes clear is how deeply unimaginative and unfunny it is. You also get the distinct sense that none of these creators have the ability to imagine things beyond “what if this dead celebrity did some buckwild shit that would have given their agents heart attacks?”
The substance of these videos speaks volumes about the current state of gen AI. But it says even more about how this technology’s output has been influenced by the gradual death of monoculture.
Though some have argued that society felt more cohesive when everyone watched the same TV shows and films — the mythical work watercooler conversation — monoculture was not without its drawbacks. That was a time when the pop cultural decision-making power was concentrated within a select pool of — typically — old, white men. Monoculture created structural barriers around the business of making art for the masses, and modern technologies like the internet and social media gave people a way to work around those gatekeepers.
It’s not a coincidence that many gen AI founders have leaned heavily into the idea that their products are designed to empower people and “democratize” the creation of art. That was the promise anyway. But when you scroll through the Sora app and see dozens of videos iterating on the same basic prompts like “celebrity or animal has been pulled over by the police under suspicion of drunk driving,” it’s hard not to see the platform as a place where users are encouraged to double down on familiar archetypes instead of making something truly original, or even remotely interesting.
Where is the “good” gen AI content, exactly?
... continue reading