It’s no secret that, despite significant investment from companies like OpenAI and Runway, AI-generated videos sometimes still struggle to be convincingly realistic. Some of the most amusing fails end up on social media, which has led to a new reaction trend on Chinese social media platforms TikTok and Bilibili, where users are creating videos poking fun at the imperfections of AI-generated content. The trend has now spread to X (formerly Twitter) in the US, where users are sharing the humorous parodies.
In particular, the videos appear to parody image synthesis videos, in which subjects seamlessly transform into other people or objects in unexpected and physically impossible ways. Chinese social media replicates these unusual visual non sequiturs without special effects, positioning their bodies in unusual ways while new and unexpected objects appear out of frame on camera.
This exaggerated imitation has struck a chord with X viewers, who find the parodies entertaining. User @theGioM shared a video, which can be seen above.”“This is high-level performance art,” wrote one X user. “Art imitating life imitating AI. I almost cried.” Another commented: “I think the only thing missing is a motorcycle that turns into a speedboat and takes off into the sky. Other than that, excellent work.”
An example of a Chinese social media video in which two people imitate imperfect AI-generated video output.
While these parodies poke fun at the current limitations, tech companies are actively trying to overcome them with more training data (examples analyzed by AI models that teach them how to create videos) and computational training time. OpenAI introduced Sora in February, which can create realistic scenes if they closely match the examples in the training data. Runways Gen-3 Alpha suffers a similar fate: It can create short clips of compelling videos within a tight set of constraints. This means that generated videos of situations outside the dataset often end up hilariously comical.
An AI-generated video in which people and animals transform in impossible ways. Social media users imitate this style.
It’s worth noting that in February, actor Will Smith beat Chinese social media users to the trend by poking fun at a terrible viral AI-generated video from 2023 of him trying to eat spaghetti. This might also bring back memories of other amusing video synthesis fails, such as the May 2023 AI-generated beer commercial created using Runway’s earlier Gen 2 model.
An example of a Chinese social media video in which two people imitate imperfect AI-generated video output.
While it may seem strange to some to imitate videos of imperfect AI, people on TikTok regularly make money by posing as NPCs (non-player characters – a term for computer-controlled video game characters).
For anyone who lived in the 1980s, the sight of this rapidly changing and often bizarre new media world can induce cognitive whiplash, but the world is a strange place, full of wonders beyond imagination. “There are more things between heaven and earth, Horatio, than thy school wisdom dreams of,” Hamlet once famously said. “Including people pretending to be video game characters and faulty video synthesis output.”
