{
  "blyg": "0.3",
  "id": "47gqe5tkz7k0dj0bn2aasvtx93",
  "kind": "fragment",
  "origin": "https://www.msweet.net/notes/",
  "page": "636-slate-crows",
  "author": {
    "name": "Matthew McDowell-Sweet",
    "url": "https://www.msweet.net/"
  },
  "created": "2026-08-23T12:36:23Z",
  "updated": "2026-08-23T12:36:23Z",
  "version": 1,
  "content_md": "We’re entering the world of post-understanding.\n\n> That sets up the question Tim keeps pressing: if an agent can produce a chip design, a proof, or a working program, how do you actually know it is correct? Passing 70% of tests is not the same as being right, and a single fabricated bug can cost a company a fortune. They dig into ProgramBench (rebuild a program from its tests, roughly 0% success), the difference between structure and competence, and the \"understanding debt\" you take on when nobody reads the code.\n\n[The Thermodynamic AI Chip · Thomas Ahle - YouTube ↗](https://youtube.com/watch?is=Bgd96nMtLNLeDkmR&v=5pieVHmlbyk)",
  "content_html": "<p>We’re entering the world of post-understanding.</p>\n<blockquote><p>That sets up the question Tim keeps pressing: if an agent can produce a chip design, a proof, or a working program, how do you actually know it is correct? Passing 70% of tests is not the same as being right, and a single fabricated bug can cost a company a fortune. They dig into ProgramBench (rebuild a program from its tests, roughly 0% success), the difference between structure and competence, and the &quot;understanding debt&quot; you take on when nobody reads the code.</p></blockquote>\n<p class=\"source\"><a href=\"https://youtube.com/watch?is=Bgd96nMtLNLeDkmR&amp;v=5pieVHmlbyk\">The Thermodynamic AI Chip · Thomas Ahle - YouTube ↗</a></p>",
  "content_hash": "sha256:eb6c7e12675bc41d98c8395e9cc663be58de00d1c2faa3b5ba6ebf7ffeccf55d",
  "media": [],
  "changelog": [
    {
      "version": 1,
      "at": "2026-08-23T12:36:23Z",
      "note": null
    }
  ]
}