AI slop is flooding scientific publishing

IEEE Spectrum has highlighted the growing problem of AI-generated slop in academic publishing through the experience of an author who used AI to produce a fake scientific paper. The account shows how quickly mass-produced, low-quality content can be created and how difficult it is for journals and peer reviewers to keep up. The experiment is not just a curiosity. It illustrates a broader strain on scientific publishing, where editors are already handling high submission volumes and reviewers are donating limited time. When convincing-looking papers can be generated at scale, the risk of wasted review effort, polluted literature, and weakened trust increases. The article notes estimates suggesting that a large share of published articles show signs of AI involvement, though such signals can be ambiguous and do not always indicate misconduct. The fake-paper experiment underscores why detection tools, clear disclosure rules, and stronger editorial screening are becoming more important. Journals need ways to verify authors, check data and citations, and identify papers that are plausible in style but thin in substance. At the same time, screening cannot rely on AI detectors alone, since they can produce false positives and miss carefully edited text. The deeper challenge is structural. Academic incentives reward publication volume, and AI makes it cheaper to produce content that looks like scholarship without adding real evidence or insight. Publishers may need to tighten submission requirements, require more detailed methods and data availability, and invest in human editorial judgment where it matters most. The goal is not to ban AI tools outright, but to prevent them from becoming a shortcut that undermines the record of scientific knowledge. If journals cannot adapt, the cost will be borne by readers, reviewers, and the credibility of research itself.