{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreicgtzizuapqbxd4m46qh3qhftjfsickxjxpmwllub3ulztie22zum",
"uri": "at://did:plc:pi6woz4d47bkuws673w2il2r/app.bsky.feed.post/3mkscqmmjuom2"
},
"path": "/t/anti-llm-sentiment-considered-harmful/14008?page=2#post_23",
"publishedAt": "2026-05-01T13:59:35.000Z",
"site": "https://discourse.haskell.org",
"textContent": "Ambrose:\n\n> to be superficially very competent but not someone I would hire - good coworkers/employees are people who don’t need that level of focus from me.\n\nThis is why I find the LLM to developer comparison ridiculous - the results depend way too much on whoever is using it.\n\nIf you had a colleague who could go from competent to unhinged because of a single word you used, that wouldn’t be a very good colleague at all.\n\nHonestly, this whole discussion is a bit weird.\n\nIf you can use an LLM and produce output which is of high enough quality that people couldn’t tell is generated (spoiler: you would have to know what you’re doing), then go ahead, _even the AI skeptics won’t mind_. If you’re hoping to prompt “make dis betr” and push a +2000 line PR expecting someone else will review it and clean it up for you, then you’re being ridiculous.",
"title": "Anti-LLM Sentiment Considered Harmful"
}