About the claim that an "AI" model could just copy itself on other machines:
These big models used in ChatGPT, Claude, etc have hundreds of billions of parameters. Even assuming that each parameter is a single 64bit floating point value (8 bytes), a 200 billion parameter model (a very conservative estimate, GPT-3 had up to 175 billion parameters 6 years ago) would be 1600TB in size.
Yeah, that will certainly not copy itself on to your laptop or phone. The claim that it could is utter nonsense.
Æ.
npub187…j005y
2026-09-15 07:57:24 UTC
Author Public Key
npub1875ph23m505q8a7ukacmm56g4h9605k0pwsuagf3qqdmnn3kqnmq4j005ySeen on
wss://relay.momostr.pinkPublished at
2026-09-15 07:57:24 UTCEvent JSON
{
"id": "50bddaea0d18dd4ea3726e27413310e4e908b759d8ab6d806a2ed99baf7db671",
"pubkey": "3fa81baa3ba3e803f7dcb771bdd348adcba7d2cf0ba1cea131001bb9ce3604f6",
"created_at": 1789459044,
"kind": 1,
"tags": [
[
"proxy",
"https://wandering.shop/@aesthr/117273987965750140",
"web"
],
[
"proxy",
"https://wandering.shop/users/aesthr/statuses/117273987965750140",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://wandering.shop/users/aesthr/statuses/117273987965750140",
"pink.momostr"
],
[
"-"
]
],
"content": "About the claim that an \"AI\" model could just copy itself on other machines:\n\nThese big models used in ChatGPT, Claude, etc have hundreds of billions of parameters. Even assuming that each parameter is a single 64bit floating point value (8 bytes), a 200 billion parameter model (a very conservative estimate, GPT-3 had up to 175 billion parameters 6 years ago) would be 1600TB in size.\n\nYeah, that will certainly not copy itself on to your laptop or phone. The claim that it could is utter nonsense.",
"sig": "a7debde14c7e67dd12309a3372e79c0fdac71a40a2d9bd7a0455f35f7b898c666a9f47dee92c8edb82c9929fd35d157c37aaf21145c19d24328f070b2b534e4c"
}