（来源：上观新闻） "AI泔水"指由生成式人工智能批量生产的低质、冗余、缺乏认知价值的内容，其以技术效率渗透数字空 间，形成了标志性的数字文化症候群，为治理数字时代低质内容提供了关键参照样本。作为数字内容生 态的"污染物"，"AI泔水"不仅挤占优质信息的传播通道，更消解内容生产的原创精神，侵蚀受众的信息 辨别能力与认知体系，对数字文化生态的健康发展构成威胁。剖析其生成机理，完善治理体系，已成为 数字时代维护信息质量、保障文化生态健康发展的关键议题。 的"AI幻觉"也会进一步加剧低质内容的传播与扩散，使数字内容生态陷入"低俗化螺旋"，进而对数字内 容生态产生持续性的破坏，增加治理的难度。 随着AI技术持续发展，"AI泔水"的形态可能不断演变，从当前的文本、图像向视频、交互式内容延伸， 治理难度也将同步提升。对此，亟需立足"法律保障、技术规制、平台责任、主体赋能"的多元协同治理 框架，完善治理路径，实现对"AI泔水"的精准治理与源头管控。 一是完善法治保障，明确治理权责边界。强化法治引领，赋能AI健康发展，加快修订数字内容治理相 关法律法规，将AI内容生成纳入法治化轨道，清晰界定"AI泔水"范畴，厘清技术企业、平 ...

Core Viewpoint - The article discusses the proliferation of low-quality AI-generated content, termed "AI Slop," which is increasingly prevalent in various sectors, including health, finance, and education. This content often lacks factual verification and can mislead readers, leading to a decline in critical thinking and judgment skills among consumers of such information [3][4][6]. Group 1: Prevalence of AI Slop - Over 38% of so-called "popular science content" on the Chinese internet is entirely AI-generated and lacks any factual verification, with this figure rising to 45% in specific fields like health and finance [4]. - The rapid production and distribution of AI-generated content are overwhelming users, with many articles containing sensational headlines but lacking substantive information [6][7]. Group 2: Characteristics of AI Slop - AI Slop is characterized by information that is hollow, emotionally manipulative, and factually ambiguous, often using vague references to studies or experts without providing specific details [7][8]. - The content is produced in bulk using templates, allowing for the generation of hundreds of articles quickly, which often results in repetitive and low-quality information [8]. Group 3: Long-term Effects of Consuming AI Slop - Prolonged exposure to AI Slop can lead to a decline in judgment and critical thinking abilities, as individuals may struggle to differentiate between true and false information [13][14]. - The impact on youth is particularly concerning, as they are in formative stages of cognitive development and may be misled by distorted facts and flawed learning methods presented in AI-generated content [15]. Group 4: Underlying Industry Dynamics - The rise of AI Slop is supported by a complete profit-driven ecosystem, where anyone with basic technical knowledge can access advanced AI tools to create content [19]. - The production of AI content is often automated, with systems capable of generating articles, images, and managing social media interactions without human oversight, leading to significant financial gains for content creators [19][20]. Group 5: Identifying and Avoiding AI Slop - To combat the spread of AI Slop, individuals are encouraged to verify sources, check for logical consistency, and be wary of sensational language commonly used in AI-generated content [21][22]. - Establishing healthy information consumption habits, such as seeking reputable sources and engaging in deep reading, is essential to mitigate the effects of low-quality content [27][28].

"AI泔水"泛滥，不利己亦不利他。从个人角度出发，直接使用AI生成的内容用于学习、科研、工作，不仅是在制造"泔水"，还 会导致自己的认知"外包"。最近一段时间，从小学生用AI写作文、大学生写论文，到基层单位借助AI撰写材料，这些事情让大 家开始担心深度思考的消逝。 "AI泔水"的影响远不止于个体。在企业中，用人工智能生产出来的"工作垃圾"也正在成为一种隐秘的负担。麻省理工学院媒体 实验室的一份报告显示，95%的组织在AI技术上的投资并未获得可衡量的回报。也有调研显示，员工每次处理"工作垃圾"平均 要花费1小时56分钟。简单来说，当某位员工借AI快速产出"看似合格"的成果，往往意味着同事需花费更多时间解读、修正，甚 至重做。除此之外，"AI泔水"也在推高整个社会的信任成本。当人们拿着AI生成的作品去发布不实信息、参评绘画展摄影展， 人与人之间的信任基础也将动摇。 若将视野放至整个数字生态，现状似乎更令人担忧。人们利用人工智能生产的作品，可能也在被另一批AI抓取、学习，最后形 成了用"泔水"制造"泔水"的恶性循环。 新闻两点论 徐也晴 最近《韦氏词典》把"slop"（意为通常由人工智能批量生成的低质量数字内容）列 ...