<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Dream-RSI on SkySeraph</title><link>https://skyseraph.github.io/ainewstags/dream-rsi/</link><description>Recent content in Dream-RSI on SkySeraph</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Sun, 27 Sep 2026 08:00:00 +0800</lastBuildDate><atom:link href="https://skyseraph.github.io/ainewstags/dream-rsi/index.xml" rel="self" type="application/rss+xml"/><item><title>Google Dream-RSI：用回放替代重跑，Agent调用减少162倍</title><link>https://skyseraph.github.io/ainews/2026/dream-rsi-recursive-self-improvement/</link><pubDate>Sun, 27 Sep 2026 08:00:00 +0800</pubDate><guid>https://skyseraph.github.io/ainews/2026/dream-rsi-recursive-self-improvement/</guid><description>Google 研究团队提出 Dream-RSI 框架，通过演化世界模型实现递归自我改进，将发现代理的调用次数减少 162 倍，为 AI 自我迭代能力提供高效实现路径。</description></item></channel></rss>