AI Slop
AI slop is low-quality, mass-produced AI-generated content published with little review: filler articles, spam images and video, fake product pages, and generated code or bug reports sent to maintainers. Wikipedia compares the term to spam; the cost-to-filter, search-degradation and maintenance effects below are the author’s characterisation, not measured here.
Why it matters
- Trust and search: authentic and generated material can be hard to tell apart (author’s view); platform labelling and demotion policies vary and were not surveyed.
- Open-source maintainers are described as receiving low-quality AI-generated issues and pull requests (no source cited here; vibe-coding without review is one possible source).
- Training data: slop re-entering web-crawled data may raise model collapse risk (hypothesis; not sourced here).
- Counter-practice: human review, provenance, fact checking, evals for content quality (llm-as-judge-and-evals).
Related: diffusion-models, ethics-in-ai, ai-literacy-and-education.
Sources
- https://en.wikipedia.org/wiki/AI_slop (secondary, fetched 2026-10-05; definition re-checked via the Wikipedia REST summary 2026-10-07: content made with generative AI perceived as lacking effort, quality or meaning, produced in volume; the term is compared to spam)
Open items
- No dated primary figures or platform policies cited; add from named sources if needed.