Eddie Roosenmaallen on Nostr: This is a very good read about how LLMs actually work, how they appear ...
This is a very good read about how LLMs actually work, how they appear “intelligent”, and how were being manipulated by the folks shilling them.
It also proposes a more correct term for what they do, it’s not “hallucinating” because there is simply no inner world nor perception; it’s better called “bullshitting”, because it has no relationship to the world or to the truth; it just sounds plausible and sometimes happens to also be correct.
https://www.scientificamerican.com/article/chatgpt-isnt-hallucinating-its-bullshitting/Published at
2024-07-19 11:53:10Event JSON
{
"id": "b39cc9b17713b9fe7b5ae1b915f610658f4975735ce12e48bc2b1c719b152631",
"pubkey": "64068411fe18523be7f95144bb5b6653890de4724fc8ac1279ce5100beed44f5",
"created_at": 1721389990,
"kind": 1,
"tags": [
[
"proxy",
"https://strangeobject.space/users/silvermoon82/statuses/112813014446222130",
"activitypub"
]
],
"content": "This is a very good read about how LLMs actually work, how they appear “intelligent”, and how were being manipulated by the folks shilling them. \nIt also proposes a more correct term for what they do, it’s not “hallucinating” because there is simply no inner world nor perception; it’s better called “bullshitting”, because it has no relationship to the world or to the truth; it just sounds plausible and sometimes happens to also be correct. \n\nhttps://www.scientificamerican.com/article/chatgpt-isnt-hallucinating-its-bullshitting/",
"sig": "70ab00f4479af412289274826c363a773d0d89d8900b721aecf164d080e7c88b45ce2bf1d7b2a90418c1752909276e6996c6a288f9a771be1ec3c5ade0616fba"
}