Greg Slepak on Nostr: Why is it that in all (?) models, the chain-of-thought is disconnected from the ...
Why is it that in all (?) models, the chain-of-thought is disconnected from the response?
I.e. if you ask some questions it's clear that what the AI "thinks" and says are two disjoint things. Sometimes one will say one thing while the other will say something totally different?
Published at
2025-02-26 03:52:41Event JSON
{
"id": "f011e5c1c4b06239840781beab401fc9a6721ff5943432fb71e6f67a37073680",
"pubkey": "fbd55141a622c22880be9a24fd1cd65d54cf4adf67cef227744d59cb24692f91",
"created_at": 1740541961,
"kind": 1,
"tags": [
[
"proxy",
"https://crib.social/objects/fc001be6-df4a-41a9-860b-ff4489f5d497",
"activitypub"
]
],
"content": "Why is it that in all (?) models, the chain-of-thought is disconnected from the response?\n\nI.e. if you ask some questions it's clear that what the AI \"thinks\" and says are two disjoint things. Sometimes one will say one thing while the other will say something totally different?",
"sig": "50f2327ab20c0845e8e862b16d0d522b7d6bf96eab158d0362ca3205248ab64a517acf0e8d9ad809b9e1ea3da259f3fe1ee8eef857c32a8297a0e42df5c60977"
}
