<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>大模型执行力 on AI内参</title>
    <link>https://www.neican.ai/tags/%E5%A4%A7%E6%A8%A1%E5%9E%8B%E6%89%A7%E8%A1%8C%E5%8A%9B/</link>
    <description>Recent content in 大模型执行力 on AI内参</description>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Mon, 13 Oct 2025 19:10:10 +0800</lastBuildDate>
    <atom:link href="https://www.neican.ai/tags/%E5%A4%A7%E6%A8%A1%E5%9E%8B%E6%89%A7%E8%A1%8C%E5%8A%9B/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>解构大模型“翻车”黑箱：从推理幻觉到执行困境，AI Agent可靠性的核心突破</title>
      <link>https://www.neican.ai/insights/ai-agent-20251013191010427-2/</link>
      <pubDate>Mon, 13 Oct 2025 19:10:10 +0800</pubDate>
      <guid>https://www.neican.ai/insights/ai-agent-20251013191010427-2/</guid>
      <description>剑桥大学等机构的最新研究颠覆了对大模型“翻车”的传统认知，指出核心问题在于其执行多步任务时的稳定性不足，而非推理能力。特别是一种名为“自条件化效应”的机制，导致模型因自身错误而持续犯错。通过采用“思考”机制和扩大模型规模，AI有望显著提升其长视界执行能力，为构建更可靠、更具经济价值的AI Agent提供了清晰的技术方向。</description>
    </item>
  </channel>
</rss>
