{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreietn2ad2gafdwkagali22r77eaitosd3peik47fv73uztpfdnd3e4",
    "uri": "at://did:plc:pi6woz4d47bkuws673w2il2r/app.bsky.feed.post/3mh4j4qsi6bc2"
  },
  "path": "/t/yet-another-opinion-on-llms-hasufells-blog/13775?page=2#post_34",
  "publishedAt": "2026-03-15T16:04:36.000Z",
  "site": "https://discourse.haskell.org",
  "textContent": "My approach is to avoid all anthropomorphic language, as Dijkstra warned. Just substitute “AI” with “computer” (or another piece of technology, like GHC) and the silliness is highlighted directly, e.g. “I asked the computer”, don’t say “it wants to” or “it makes mistakes” — a computer program does what it’s programmed to. In the engineering world we don’t say that software “makes mistakes”, we call those errors or failures, or some other technical language. No one says that a Bayesian spam filter “makes mistakes” if some spam gets through.\n\nAnthropomorphic language is fun and whimsical in a Terry Pratchett kind of way, but only if it’s self-aware, or tongue-in-cheek. I think the problem with LLMs and “AI” is that people seem to forget they’re doing it, and it maybe reflects deeper perceptions.",
  "title": "Yet another opinion on LLMs · Hasufell's blog"
}