AI 'Slop' Crisis Overwhelms Computer Science

The article highlights a growing crisis in computer science caused by the proliferation of AI-generated research papers, often referred to as 'AI slop.' The ease with which LLMs can generate hypotheses, write code, and draft papers has led to a surge in submissions to conferences and preprint repositories, overwhelming the existing peer review system. The International Conference on Machine Learning (ICML) has experienced a dramatic increase in submissions, more than doubling from 2025 to 2026. This surge is attributed to the increased productivity of researchers using LLMs, with some studies suggesting productivity gains of up to 89.3%. However, the increased volume has made thorough and careful evaluation increasingly difficult, leading to concerns about the quality and validity of published research. Many authors are failing to properly validate or verify AI-generated content, resulting in the publication of fake or low-quality papers. The arXiv preprint repository has also seen a significant increase in submissions and rejections since the advent of ChatGPT. In response to this crisis, various measures are being implemented, including the use of AI in peer review, eligibility checks for first-time submitters, and submission fees. These measures aim to combat the spread of AI slop and maintain the integrity of scientific research. The stakes are high, as the erosion of trust in computer science research could have significant consequences for the field and for society as a whole.