mleku on Nostr: seems like a lot of people on nostr are in love with GPTs and stabledifusions but ...
seems like a lot of people on nostr are in love with GPTs and stabledifusions but there is some of us old farts who are like "not this clippy shit again" and talking about how it's ruining programming
not a luddite
GPTs simply don't have the causality analysis capability to do real programming
yes, they can build front ends with them but that's a layer that "presentation devices" basically dumb GUI terminals using RDP and X) have traditionally been called "thin clients"
the AIs can't do advanced, let alone concurrent servers, and they are no good at even understanding or generating actually correct or compilable, even, code in systems programming languages
they don't even get simple things like static typing or interfaces or objects right, because they require deep (ie, multiple levels of causality) thinking.
even the smartest humans struggle to do deep thinking without taking a long time, and even then they will mess things up.
i'd be willing to bet you can take a 5 year old child and teach them to build web app front ends. that's about the intelligence level we are talking.
i know that as a 9 year old i was capable of linguistic cognition at what was average for humans at 15, and i'd guess that at best these AIs are going to give us plausible simulacra of the thinking ability of domestic pets.
Published at
2025-06-01 18:18:53Event JSON
{
"id": "474f7819853639bd5b9704fdbcda0d4fa4e079f3d5258caeae0cd419c3240e5e",
"pubkey": "4c800257a588a82849d049817c2bdaad984b25a45ad9f6dad66e47d3b47e3b2f",
"created_at": 1748801933,
"kind": 1,
"tags": [
[
"e",
"8cc7d6337da8d6192a590713e76425427a78ac52c85c39e0e082b72433dba035",
"wss://theforest.nostr1.com/",
"root",
"460c25e682fda7832b52d1f22d3d22b3176d972f60dcdc3212ed8c92ef85065c"
],
[
"p",
"edb470271297ac5a61f277f3cd14de54c67eb5ccd20ef0d9df29be18685bb004"
],
[
"p",
"460c25e682fda7832b52d1f22d3d22b3176d972f60dcdc3212ed8c92ef85065c"
]
],
"content": "seems like a lot of people on nostr are in love with GPTs and stabledifusions but there is some of us old farts who are like \"not this clippy shit again\" and talking about how it's ruining programming\n\nnot a luddite\n\nGPTs simply don't have the causality analysis capability to do real programming\n\nyes, they can build front ends with them but that's a layer that \"presentation devices\" basically dumb GUI terminals using RDP and X) have traditionally been called \"thin clients\"\n\nthe AIs can't do advanced, let alone concurrent servers, and they are no good at even understanding or generating actually correct or compilable, even, code in systems programming languages\n\nthey don't even get simple things like static typing or interfaces or objects right, because they require deep (ie, multiple levels of causality) thinking.\n\neven the smartest humans struggle to do deep thinking without taking a long time, and even then they will mess things up.\n\ni'd be willing to bet you can take a 5 year old child and teach them to build web app front ends. that's about the intelligence level we are talking.\n\ni know that as a 9 year old i was capable of linguistic cognition at what was average for humans at 15, and i'd guess that at best these AIs are going to give us plausible simulacra of the thinking ability of domestic pets.",
"sig": "098f26d64b71262a3a57027564e404aa029162817c42179f9d5d8f0846339f789b80abb0a5a9c3de0cb2ada5fcaa529907d84ca8bf084b0de9a3f8560a8f762e"
}