AI-generated video is having a bit of a moment right now, as many AI enthusiasts are experimenting with the technology using popular meme templates.
AI-generated video has come a long way since those clips of Will Smith eating spaghetti went viral; KwaiCut, a Chinese AI video generation app was used to create a clip of a man eating noodles, and it was perfectly passable (even if his fingers are a bit wonky).
OpenAI’s Sora has released a bunch of curated demo footage, some of which is quite impressive, but the model hasn’t been released to the public yet.
Luma Labs beat them to the punch, having released an AI video generator dubbed “Dream Machine,” which is free to try. Like AI image generators, it takes few prompts to produce something worth sharing, and many posted their best results online.
Is AI-Generated Video Worth Watching?
While AI-generated video has dramatically improved, and will continue to improve, even the strongest results play like a fever dream, because the people depicted in the videos tend to shapeshift; as is the case with all generative AI models, so-called “hallucinations” are unavoidable.
Replicating human movement is no easy feat, even for seasoned animators, and automating the process produces some very uncanny results.
One of the most unsettling clips emerged from the “distracted boyfriend” meme, as said boyfriend begins to follow the woman who distracts him while she slowly transforms into a child, for some reason.
The meme’s cousin, the “distracted girlfriend,” produced an even more jarring video, with said girlfriend’s body twisting as though she were in the midst of an exorcism — definitely the stuff of nightmares.
A common reaction GIF featuring an exasperated Captain Picard (Patrick Stewart) had something of a plot twist, revealing a completely different face for the character, invented by AI.
One of the most interesting side-effects of “hallucinations” is the creation of non-existent people; for example, converting Ellen DeGeneres’s 2014 viral Oscars selfie to video reveals an unknown celebrity wearing shades, an AI-invention who seems somewhat surprised to be there.
A video version of the iconic album cover for Abbey Road even generates a “fifth Beatle” to join the others.
Amusingly, this isn’t the first time that an AI model has generated a fifth member of The Beatles.
While AI enthusiasts appeared impressed by the animated abominations, critics mocked the videos, pointing out that generative AI is guzzling an astonishing amount of water and electricity to create unintentionally eerie videos.
Considering how extraordinarily difficult it is for a professional team of filmmakers to conceive, stage and shoot a single compelling scene, it seems unlikely that generative AI will ever produce results that can truly compete with human-crafted work.
The technology has no capacity to understand context, emotion, nuance, or perspective — if there’s any artistic integrity to any of this, it might just be in the realm of body horror.
