<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai Code Validation on No Semicolons</title><link>https://nosemicolons.com/tags/ai-code-validation/</link><description>Recent content in Ai Code Validation on No Semicolons</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 15 Sep 2026 13:00:54 +0000</lastBuildDate><atom:link href="https://nosemicolons.com/tags/ai-code-validation/index.xml" rel="self" type="application/rss+xml"/><item><title>The AI Code Generation Model Consensus Crisis: How 3-Model Validation Reduced My Bug Count by 85%</title><link>https://nosemicolons.com/posts/ai-code-generation-model-consensus-validation/</link><pubDate>Tue, 15 Sep 2026 13:00:54 +0000</pubDate><guid>https://nosemicolons.com/posts/ai-code-generation-model-consensus-validation/</guid><description>&lt;p>What if I told you that the AI model you&amp;rsquo;re using to generate code is probably wrong about 30% of the time, but you could catch most of those errors before they ever see your codebase?&lt;/p>
&lt;p>Three months ago, I was that developer who&amp;rsquo;d copy-paste Claude&amp;rsquo;s output straight into my editor, run a quick test, and call it done. My pull requests were getting rejected left and right for subtle bugs that somehow slipped through my testing. Edge cases I hadn&amp;rsquo;t considered. Logic errors that looked right at first glance but fell apart under scrutiny.&lt;/p></description></item></channel></rss>