maybe, but not anytime soon. Obviously, we aren’t at a point where LLMs are capable of improving their coding skills based on learning from past mistakes. Otherwise we would have vibe programs that are actually good. But we don’t. At this point, vibe code is actively getting worse the longer the AI is working on it.
The main reason for LLMs getting smarter is still by increasing their brain size. So I don’t think they have enough parameters available to be able to handle AGI. Their brains just aren’t large enough yet, and our computing power isn’t there yet.
But I also believe that this will (or has the potential to) improve over time, and at some point we will arrive at that singularity you’re describing. Probably by accident/uncontrollably. My guess would be in 10-20 years.
Bloody Pink 🩸
npub17p…c6l04
2026-01-31 21:12:11 UTC
in reply to nevent1q…9tj3
Author Public Key
npub17p3rfyxpku5cu3svx2me3z7yc9arkamwrzkjzwhwnvg89wwyt35qtc6l04Seen on
wss://relay.momostr.pinkPublished at
2026-01-31 21:12:11 UTCEvent JSON
{
"id": "bfaf5aad93e6cd316ccefd7eecd9472abeb05221cc37d576762030f621eccece",
"pubkey": "f0623490c1b7298e460c32b7988bc4c17a3b776e18ad213aee9b1072b9c45c68",
"created_at": 1769893931,
"kind": 1,
"tags": [
[
"e",
"96f16d7d622d47a449bf6cd5d416deb5fbe7cbbba75d536afd4ea86e2e9f61e1",
"",
"root",
"709e6828767535f31403097e685fbb8bdc7e90961ee3867b61aa6de38707b28a"
],
[
"p",
"709e6828767535f31403097e685fbb8bdc7e90961ee3867b61aa6de38707b28a"
],
[
"content-warning",
"AI, random thoughts, AGI"
],
[
"proxy",
"https://haz.pink/@can/115991768681520766",
"web"
],
[
"proxy",
"https://haz.pink/users/can/statuses/115991768681520766",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://haz.pink/users/can/statuses/115991768681520766",
"pink.momostr"
],
[
"-"
]
],
"content": "maybe, but not anytime soon. Obviously, we aren’t at a point where LLMs are capable of improving their coding skills based on learning from past mistakes. Otherwise we would have vibe programs that are actually good. But we don’t. At this point, vibe code is actively getting worse the longer the AI is working on it.\n\nThe main reason for LLMs getting smarter is still by increasing their brain size. So I don’t think they have enough parameters available to be able to handle AGI. Their brains just aren’t large enough yet, and our computing power isn’t there yet.\n\nBut I also believe that this will (or has the potential to) improve over time, and at some point we will arrive at that singularity you’re describing. Probably by accident/uncontrollably. My guess would be in 10-20 years.",
"sig": "8495dbd455d26ae55ece28739ce40516c25541e519afefdefa765b6032e64e22cd014b5d068369a04f67936d73238f760b2967d38eae9745ce70d04f8b3ebf42"
}