What is AI Slop? An Explication

From 2024 onwards, the term ‘AI slop’ proliferated across academic disciplines and public discourse, yet it has not converged on a clear definition that scholars and practitioners can agree on, with existing characterisations differing on what renders slop defective. This paper argues that these persistent disagreements are not accidental but track the disciplinary standards authors draw on. Treating ‘slop’ as an exploratory concept, the paper first maps extant proposals onto four dimensions – medium, epistemic standing, functional role, and producers’ intentions – and demonstrates that disagreement revolves around what makes slop inadequate. This disagreement, however, masks a deeper structural convergence, reflected in a tentative definition: slop is synthetic media that presents itself as meeting a standard of adequacy which, on engagement, it manifestly fails to meet, where the requisite standard is the one that conventionally governs the kind of artefact it presents itself as being. Three such standards (epistemic, aesthetic, and communicative) are distinguished; the communicative standard is argued to be the most distinctive. The account is tested against neighbouring notions: kitsch, bullshit, fake news, deepfakes, and spam. Intentions are relegated to the level of deployment rather than constitution, resulting in a more robust account across different contexts and applications.