AI slop has become a familiar presence on social media, where it appears as overprocessed, derivative content that often feels uncanny or absurd. A new paper on arXiv investigates how ordinary users understand and talk about this phenomenon, treating 'AI slop' as a folk category that is both widely used and heavily contested.
The paper draws on examples such as an image of Jesus fused with a shrimp, a humanoid orange cat shoplifting, and a dog heroically rescuing a child after a plane crash. These cases illustrate the range of content that gets labelled as slop, from the surreal to the falsely sentimental.
By examining how the term is deployed, the authors show that 'AI slop' is not a neutral description but a way of marking content as defective, amplified, or contested. The study contributes to ongoing discussions about how audiences make sense of generative AI's impact on online spaces.