<?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>Scientific Discovery on Shuibai’s Blog</title>
    <link>https://zhangshuibai.github.io/blog/tags/scientific-discovery/</link>
    <description>Recent content in Scientific Discovery on Shuibai’s Blog</description>
    <generator>Hugo -- 0.150.0</generator>
    <language>en</language>
    <lastBuildDate>Sat, 19 Sep 2026 12:00:00 -0500</lastBuildDate>
    <atom:link href="https://zhangshuibai.github.io/blog/tags/scientific-discovery/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Scientific discovery is the next bottleneck for intelligence</title>
      <link>https://zhangshuibai.github.io/blog/posts/scientific-discovery-is-the-next-bottleneck-for-intelligence/</link>
      <pubDate>Sat, 19 Sep 2026 12:00:00 -0500</pubDate>
      <guid>https://zhangshuibai.github.io/blog/posts/scientific-discovery-is-the-next-bottleneck-for-intelligence/</guid>
      <description>&lt;p&gt;Post-training climbed maths, code and tool use on process data written by humans or synthesised under their direction. The research rung has no such corpus. To keep improving, models have to do science — which makes discovery AI&#39;s own bottleneck, not an application of it.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
