this is a reminder that LLMs hallucinate by default and guess correctly by luck. this is also a reminder that everything that an LLM tells you that can't be readily checked/evaluated should be treated like rubbish.
the fun thing is that there are plenty of use cases when you *can*, so they can be pretty useful! but don't ask it facts, or opinions, or choices.
Matteo Capucci
npub1xv…58vg4
2026-02-03 08:19:03 UTC
in reply to nevent1q…x32a
Author Public Key
npub1xvgrm0sxezg5zejxd8hgtrpmnkp4a6swzt6td4tmx08tnlss82hqm58vg4Seen on
wss://relay.ditto.pubPublished at
2026-02-03 08:19:03 UTCEvent JSON
{
"id": "204cccc269abf247c6f5e6993fc93d1cf506e366e67b0f5f4728cb274eaa0afa",
"pubkey": "33103dbe06c89141664669ee858c3b9d835eea0e12f4b6d57b33ceb9fe103aae",
"created_at": 1770106743,
"kind": 1,
"tags": [
[
"e",
"d1c0c6f29cb9f4d4d750d0b162992d08973111cda390ae91f78540ec0b2047f5",
"wss://relay.ditto.pub",
"reply"
],
[
"proxy",
"https://mathstodon.xyz/users/mc/statuses/116005715541665664",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.ditto.pub"
]
],
"content": "this is a reminder that LLMs hallucinate by default and guess correctly by luck. this is also a reminder that everything that an LLM tells you that can't be readily checked/evaluated should be treated like rubbish.\n\nthe fun thing is that there are plenty of use cases when you *can*, so they can be pretty useful! but don't ask it facts, or opinions, or choices.",
"sig": "b5cbe532110408e24d3a796bcf402c9188de82773acf3e11f5863df0b5d1a1278bd1872898b33e8b91cb07ab2ccb65113a06825e13d216bd1fd37ff652ccb3be"
}