Generative AI threatens to disrupt how we work, learn, communicate, and participate in online communities. We performed a qualitative interview study to understand how moderators of online communities on the social sharing site Reddit are challenged by AI-generated content (AIGC) and how they are adapting. We conducted fifteen in-depth, semi-structured interviews with subreddit moderators about their experiences with AIGC. Though our participants see both legitimate and illegitimate motivations for using AIGC, on the whole they view it as detrimental to their communities, with a level of concern that is dependent on the purpose and size of their subreddits. Moderators reported developing rules and using a variety of strategies that may help communities prevent or curb AIGC, but without foolproof detection tools, enforcement is challenging and relies on heuristics. AIGC poses a unique threat to online communities who value authenticity, in that even well-intentioned community members may inadvertently cause disruption. Overall, for online communities, the threat of Generative AI is not speculative: the disruption has already begun.

翻译：生成式AI有可能颠覆我们工作、学习、交流以及参与在线社区的方式。我们开展了一项定性访谈研究，以了解社交分享网站Reddit上在线社区版主如何应对AI生成内容（AIGC）带来的挑战，以及他们正在如何适应。我们与子版块版主进行了十五次深入的半结构化访谈，了解他们与AIGC相关的经历。尽管受访者认为使用AIGC既有合理动机也有不当动机，但总体而言，他们视其对其社区有害，并且担忧程度取决于其子版块的目的与规模。版主报告称，他们制定了规则并采用了多种策略，可能有助于社区预防或遏制AIGC，但由于缺乏万无一失的检测工具，执行工作仍面临挑战，且依赖于经验法则。AIGC对重视真实性的在线社区构成了独特威胁，即使是出于善意的社区成员也可能无意中造成干扰。总体而言，对在线社区而言，生成式AI的威胁并非臆测：颠覆已经开始。