创建

登录 ReadmeX

或

了解我们如何处理你的信息: 隐私政策

ReadmeX

陪你读懂 AI 动态

今天,想先了解什么?

选一份动态总结,或直接聊聊你的问题。

社区和关注动态总结只属于你自己。

资讯

纳德拉称模型扩展不会自动带来对齐

AI 总结

微软CEO萨提亚·纳德拉表示,他已不再相信提升模型能力会自动带来AI对齐,并称这是一个更难且不同的问题。他认为,对齐主要需要在训练阶段完成,甚至可能从预训练开始,同时呼吁企业围绕AI模型和智能体部署零信任安全、不可篡改日志与审计机制。纳德拉还称思维链可能不忠实,并表示微软的MAI模型将以授权权重形式销售,而不会作为开放权重模型发布。

深度见解

纳德拉观点的转变质疑了仅靠扩展模型即可实现对齐的假设,并提出训练干预、零信任控制和透明度等配套保障方案。

Satya Nadella says he was wrong to think AI alignment would come with scale
Microsoft chairman and CEO Satya Nadella on stage at Microsoft Build 2026 in San Francisco Credit: Photo by Dan DeLong. Used with permission from Microsoft.
Image Credits
Microsoft chairman and CEO Satya Nadella on stage at Microsoft Build 2026 in San Francisco Credit: Photo by Dan DeLong. Used with permission from Microsoft.

Microsoft chief executive Satya Nadella no longer believes AI alignment will come just from making models more capable. He said so on a new episode of the 20VC podcast. Host Harry Stebbings had asked what he had changed his mind about in the past year.

#Microsoft#Microsoft Agent 365#Microsoft Perception#MAI models

原文约 3 分钟读完阅读原文
群聊

还没有评论,来说说你的看法。