A | 8 月 25 日消息，LinkedIn 表示，AI 劣质帖子的浏览量已不如从前，同时，被用户怀疑发布此类内容的人也将开始收到相关反馈。 FORT LAUDERDALE, Fla. -- A California property manager was sentenced to a year in federal prison for sending more than 200 vile online messages to a father of a teenage girl who died in the 2018 massacre at Florida's Marjory Stoneman Douglas High School. James Catalano, 62 of Fresno, was sentenced Friday by U.S. District Judge Robert Scola in Miami after pleading guilty in March to cyberstalking. Prosecutors called the messages he sent Fred Guttenberg “callous and cruel.”Guttenberg's 14-year-old daughter Jaime Guttenberg was murdered in the Feb. 14, 2018, shooting in Parkland that left 14 students and three staff members dead. Catalano also received three years probation and must undergo mental health treatment. Catalano sent Guttenberg messages for eight months starting in December 2021 that celebrated Jaime Guttenberg's death and reveled in the wounds she suffered. He also mocked the sadness and loss Guttenberg feels and directed obscenities, slurs and disturbing insults to him and his daughter. Catalano told investigators he was angry at Guttenberg for his outspoken advocacy for stronger gun laws since his daughter's death. He told them he believed Guttenberg was using his daughter's death “to push his political agenda” and was “trying to put (Guttenberg) in check by sending him the messages.”“By his own admission, the defendant was motivated to stalk the victim and send him heinous messages simply because he disagreed with the victim’s political views,” Assistant U.S. Attorney Arielle Klepach wrote in court documents. “He capitalized on the victim’s grief and the horrific nature of his daughter’s death in order to silence him.” She wrote that Catalano sent similar messages to others, but he has not been charged in those cases. Guttenberg said Monday that the sentence “is a big deal” and sends a message to those who cyberstalk the families of shooting victims that they will be caught and punished. He said Judge Scola agreed that while none of the messages contained a direct threat, in their totality they constituted one. Catalano's attorneys did not immediately return a call Monday seeking comment. The former student who murdered Jaime Guttenberg and the others is serving a life sentence.。LinkedIn 首席产品官 Hari Srinivasan 表示，与几周前相比，目前该职业社交平台认定为 AI“垃圾内容”的帖子，整体浏览量已经下降了 40%。“尽管我们已经取得了一些进展，但我们知道还有很多工作要做，以确保 LinkedIn 始终是一个能够让你找到真实的人和真实观点的地方…… 这仍然是我们非常关注的问题。”Srinivasan 在 LinkedIn 最近发布的一篇帖子中写道。Srinivasan 表示，自 LinkedIn 于 7 月 30 日推出“看起来像是 AI 垃圾”（seems like AI slop）按钮以来，已经有超过 100 万人点击该按钮，对他们认为可疑的帖子和评论进行反馈。当时，Srinivasan 强调，LinkedIn 更希望依靠用户之间的反馈，而不是使用自动化检测器来识别 AI 生成内容。
B | “我们希望会员能够从真实的人那里获得关于内容是否听起来真实的反馈，而不仅仅是让 AI 检测器进行判断，然后出现误判。”他写道。Srinivasan 表示，如果一篇帖子获得了“足够多的社区反馈”，LinkedIn 现在会在“帖子分析”（Post Analytics）中通过一条提示信息突出显示这一情况，“帮助你了解其他人如何看待你的内容”。他还表示，LinkedIn 已经“建立了相应的保护机制”，防止用户的内容遭到其他用户不公平地集中举报或针对。“我们从善意出发来设计这项机制；我知道自己越来越注意如何避免让内容听起来像 AI，而我们的目标是提供有帮助的反馈。”他写道。近期，随着 Anthropic 的 Claude、OpenAI 的 ChatGPT 等 AI 工具越来越普及，包括 Substack 和 YouTube 在内的其他平台也开始进一步收紧对 AI 生成内容的态度。注意到，Business Insider 的 Katie Notopoulos 最近体验了一款名为 Cringebot 3000 的 AI 工具。这款工具专门模仿 LinkedIn 上无处不在的“思想领袖”式写作风格。她发现，与自己平时发布的帖子相比，这款工具生成的内容在 LinkedIn 上获得了更多阅读量。随着 LinkedIn 最新调整上线，Cringebot 以及其他 AI 聊天机器人生成的低质量内容可能不再触达那么多用户，同时也可能引发用户的相关反馈。
Current article:http://44bwo.qixicuorenchouzuan.buzz/list_08ut/a8aq0c.html
Published on:09:06:58
新闻热点
新闻爆料
图片精选
点击排行
