Reddit用户搭建众包工具，识别AI生成的社交媒体假资料，观察数月发现常见模式
英文摘要
A Reddit user documented consistent visual and behavioral patterns in AI-generated social media profiles across multiple platforms, including perfect lighting, generic bios, low following-to-follower ratios, and emoji-only comments. They noted that engagement algorithms amplify these fake profiles once a user interacts with them. In response, the user built a community tool where participants can flag and vote on suspicious profiles, leveraging human pattern recognition through crowdsourcing. The project is intended as a collective observation effort rather than an authoritative judgment system.
中文摘要
一位Reddit用户记录了多个社交媒体平台上AI生成账号的常见视觉和行为模式，如完美的面部光线、千篇一律的个人简介、关注数极少但粉丝量巨大、以及纯表情评论。他们指出，一旦用户与这些账号互动，推荐算法会加剧其传播。为此，该用户搭建了一个社区工具，允许参与者对可疑账号进行标记和投票，通过众包方式利用人类的模式识别能力。该项目旨在成为集体观察的努力，而非权威判定系统。
关键要点
User identified cross-platform patterns in AI-generated profiles: consistent lighting glows, generic bios, low following but high follower counts, and emoji-only comments.
用户发现了AI生成资料在多平台上的共有模式：持续的面部光线辉光、通用个人简介、关注少粉丝多、以及纯表情评论。
Social media algorithms were found to increase the visibility of these fake profiles once a user engages with any AI-generated content.
发现一旦用户与任何AI生成内容互动，社交媒体算法会提升这些虚假资料的曝光度。
The user created a crowdsourcing tool that enables the community to flag and vote on suspicious profiles, relying on collective human judgment.
用户搭建了一个众包工具，让社区成员对可疑资料进行标记和投票，依赖集体的人类判断。