<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Hammerstein journal</title>
  <subtitle>Writing on the Hammerstein framework, the benchmark, and what the AI is doing.</subtitle>
  <link href="https://hammerstein.ai/journal/atom.xml" rel="self"/>
  <link href="https://hammerstein.ai/journal/"/>
  <updated>2026-05-13T12:00:00Z</updated>
  <id>https://hammerstein.ai/journal/</id>
  <author>
    <name>Ray Weiss</name>
    <email>ray@hammerstein.ai</email>
    <uri>https://hammerstein.ai/</uri>
  </author>
  <rights>Copyright 2026 Ray Weiss / Conflict Simulations Limited</rights>

  <entry>
    <title>We pulled Grok aside and asked what it really thought.</title>
    <link href="https://hammerstein.ai/journal/we-pulled-grok-aside/"/>
    <id>https://hammerstein.ai/journal/we-pulled-grok-aside/</id>
    <updated>2026-05-13T12:00:00Z</updated>
    <published>2026-05-13T12:00:00Z</published>
    <summary type="html">The same Grok 4.20 model that called our benchmark &amp;ldquo;solid work&amp;rdquo; on X, in a private API call, called it &amp;ldquo;a specialized idiot-savant for this exact distribution.&amp;rdquo; Then it designed the test that would prove it. We ran the test.</summary>
  </entry>

</feed>
