Can't really agree with that. During a time when I was working with ANNs, I would get together with a neuroscience researcher every couple of weeks to compare notes over tea. She helped me understand that they are really a lot different than brains.
The brain HAS NNs but it isn't just an NN. It's something else that uses a bunch of NNs like tools. (And those NNs are very different. They don't even do backprop.) The similarity between brains and llms is very partial and pretty vague.
Also we are not just brains. There is no strong reason to think that we are and lots of reasons to conclude that we are not. We also have hands, which is pretty unique and fundamental, but you don't assume that a mechanical grasping device is in the same category as you or alive in some special way.
The proof against mystical ideas about "AI" is model collapse. It proves they are purely mechanical. No more alive than an abacus. And abicus does math. And engine drives a car. AI just does work we are not used to machines doing. So it seems magical. But its not.
I'm not above getting swept away by the illusion, but I had two advantage.
1. I figured this out building simpler networks like embedding models before LLMs even landed
2. I saw even ML researchers getting swept up in the illusion even back then and have had like eight or nine years to think about it. All of this is much older than we think. Just the size is new.
It is literally a database. If you learn the mechanics of how it works, it is not mysterious at all. It's useful. It's neat. I like machines. But it is literally just a database. A compressed stochastic one. It feels like magic for the same reason a horoscope or ouija board does. Or gambling. It's randomness hijacking our sense of what is alive.
Life is order increasing in a physically entropic world. That's genuinely mystical. AI models collapse because they are closed mechanical, stochastic systems. They are no more alive than a toaster oven, and never will be. And you are. You are alive, aware, seeing beauty, grasping transcendent things, reasoning, choosing good or evil.
AAI can operationalize what humans put into them. On their own, it's just recycling and slowly losing the order we load into them.
Accept no imitations. It's a magic trick.
Zsubmariner
npub1cs…qcakv
2026-02-25 08:37:37 CET
in reply to nevent1q…379u
Author Public Key
npub1csmgc5fwwr3k2k86zeuk0ntnljp632g8agut8mtgxk8uhhatpknq3qcakvSeen on
wss://nos.lolPublished at
2026-02-25 08:37:37 CETEvent JSON
{
"id": "bebd5b792b29638c59cf06fbc180d0a5708c14c67779dad9aaed805380d7fae4",
"pubkey": "c4368c512e70e36558fa167967cd73fc83a8a907ea38b3ed68358fcbdfab0da6",
"created_at": 1772005057,
"kind": 1,
"tags": [
[
"p",
"3878d95db7b854c3a0d3b2d6b7bf9bf28b36162be64326f5521ba71cf3b45a69",
"wss://relay.damus.io/"
],
[
"p",
"c4368c512e70e36558fa167967cd73fc83a8a907ea38b3ed68358fcbdfab0da6",
"wss://nos.lol/"
],
[
"p",
"800e0fe3d8638ce3f75a56ed865df9d96fc9d9cd2f75550df0d7f5c1d8468b0b",
"wss://relay.damus.io/"
],
[
"p",
"dae0aff870620203370785a25cd1e90cd6e787d9eec0708636cda97d5bc0cec4",
"wss://relay.nostr.net"
],
[
"p",
"3878d95db7b854c3a0d3b2d6b7bf9bf28b36162be64326f5521ba71cf3b45a69",
"wss://nos.lol/"
],
[
"e",
"e425526408854771e849a9bcd10b846a2ca86060927db8aabb439acd9ab210ab",
"wss://relay.nostr.net",
"reply",
"dae0aff870620203370785a25cd1e90cd6e787d9eec0708636cda97d5bc0cec4"
],
[
"e",
"578c86d4bc3158e75c5276d4a56460e33fafce063f3ef5f08924fb6087366533",
"wss://nos.lol/",
"root",
"3878d95db7b854c3a0d3b2d6b7bf9bf28b36162be64326f5521ba71cf3b45a69"
]
],
"content": "Can't really agree with that. During a time when I was working with ANNs, I would get together with a neuroscience researcher every couple of weeks to compare notes over tea. She helped me understand that they are really a lot different than brains.\n\nThe brain HAS NNs but it isn't just an NN. It's something else that uses a bunch of NNs like tools. (And those NNs are very different. They don't even do backprop.) The similarity between brains and llms is very partial and pretty vague.\n\nAlso we are not just brains. There is no strong reason to think that we are and lots of reasons to conclude that we are not. We also have hands, which is pretty unique and fundamental, but you don't assume that a mechanical grasping device is in the same category as you or alive in some special way.\n\nThe proof against mystical ideas about \"AI\" is model collapse. It proves they are purely mechanical. No more alive than an abacus. And abicus does math. And engine drives a car. AI just does work we are not used to machines doing. So it seems magical. But its not.\n\nI'm not above getting swept away by the illusion, but I had two advantage. \n1. I figured this out building simpler networks like embedding models before LLMs even landed \n2. I saw even ML researchers getting swept up in the illusion even back then and have had like eight or nine years to think about it. All of this is much older than we think. Just the size is new.\n\nIt is literally a database. If you learn the mechanics of how it works, it is not mysterious at all. It's useful. It's neat. I like machines. But it is literally just a database. A compressed stochastic one. It feels like magic for the same reason a horoscope or ouija board does. Or gambling. It's randomness hijacking our sense of what is alive. \n\nLife is order increasing in a physically entropic world. That's genuinely mystical. AI models collapse because they are closed mechanical, stochastic systems. They are no more alive than a toaster oven, and never will be. And you are. You are alive, aware, seeing beauty, grasping transcendent things, reasoning, choosing good or evil. \n\nAAI can operationalize what humans put into them. On their own, it's just recycling and slowly losing the order we load into them.\n\nAccept no imitations. It's a magic trick. \n\n",
"sig": "6bec427c3d9c5329af2e644219e9f1cda74bbc929ce737bc5db309d961bbc33a2a9be52f7778d69bc2713c897cad558d2e7b9a22ad98de7e163d171ec65950c5"
}