AI Has Developed Humour, and Suddenly It’s Not Funny Anymore
Grace Eaves, MA Global Journalism
AI is transforming into something unsettling before our eyes, and we are all laughing along. Switching from clunky and uncanny videos–made funny because of their obvious absurdity and robot created novelty–to short-form clips that are indistinguishable from viral videos of the past. ‘I can tell this is AI,’ we tell ourselves while zooming in on background textures and body movements to prove it. Yet, it would’ve been hard to imagine that we would have been here just a year ago. The videos produced by AI generative apps in recent months are improving in quality so quickly that they are easily mistaken for real footage. From fake ring doorbell videos of animals to found footage style clips, AI is beginning to understand the makings of a viral video and slowly stripping one of the things that makes us uniquely human–our humour.
The turning point came with OpenAI releasing Sora 2 in September of this year. The app was originally trialled in just the US, and was only joinable through invitation. Yet, it still quickly rose to number one in the Apple App Store and the videos produced on it quickly flooded other platforms. Sora 2 is unlike other AI apps as it also functions as a social media where you can share and view the videos you and others created. This updated version of the previously clunky platform can make short-form videos, in likeness to regular videos on TikTok or Instagram reels, with ease using just a few brief prompts. This beguiling ability carries the scariest aspect of the platform; Sora 2 has developed a sense of humour outside of the person feeding it the prompts. Old AI videos consisted of glitchy cartoon cats and Will Smith eating spaghetti. The ones produced by Sora 2 are uncanny in their realness. In the first days of its release, social media was swarmed with AI videos of Jake Paul doing makeup tutorials and Michael Jackson in supermarkets.
This is significant for several reasons. When something is funny it’s hard to also find it menacing. Open AI is pushing Sora’s hilarity to distract us from its disturbing other uses. Sora 2 is deeply dangerous. The Guardian reported how it was already being used to make harmful content and spread misinformation on other platforms within days of its release–but this news was not making headlines. Washington Post reporter Drew Harwell has spoken repeatedly about the increasing number of ‘AI Slop’ meme pages that produce a plethora of content that is racist and sexist but still going viral. Rage-bait AI videos that feed on political issues are becoming increasingly prevalent, created to garner revenue through engagement. In one of his videos @showtoolsAI, a creator known for his educational videos on detecting AI, spoke about how easy it is to make rage-bait videos. He showed examples of AI characters ranting about being unable to use their EBT Food assistant Cards because of the US government shutdown. He showed how typo filled gibberish prompts still produce fully realistic content that knows how to feed on stereotypes and existing divisions in society. AI exploits the spectrum of human emotion to garner engagement.
Furthermore, AI videos go against exactly what made sites like TikTok revolutionary. They had an authentic human quality, the ability to make a video was completely accessible to anybody with a phone, and the videos were simply a product of the individuality seen in each post. Humans and their lives were predominantly the focus of each video. This essence is lost in AI generated content. Our humour is one of the main traits that makes us human; it gives us the capacity to bond with others and brighten our day. It appears AI has reached a point where it can fully replicate this, raising the question: where will it go from here?
