<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Ai-Benchmark on Liuji's Blog</title><link>https://liuji.cc/tags/ai-benchmark/</link><description>Recent content in Ai-Benchmark on Liuji's Blog</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Mon, 11 May 2026 11:30:00 +0800</lastBuildDate><atom:link href="https://liuji.cc/tags/ai-benchmark/index.xml" rel="self" type="application/rss+xml"/><item><title>GPT-5.5 登顶全球 AI 榜单，但"越聪明越瞎编"成最大软肋</title><link>https://liuji.cc/posts/gpt55-hallucination/</link><pubDate>Mon, 11 May 2026 11:30:00 +0800</pubDate><guid>https://liuji.cc/posts/gpt55-hallucination/</guid><description>GPT-5.5 是目前最强的 AI，也是目前最自信的说谎者——这不是矛盾，而是当前 AI 架构的根本困境。</description></item></channel></rss>