Mark Atwood on Nostr: Tasks where a LLM can get immediate feedback from reality, and where experimentation ...
Tasks where a LLM can get immediate feedback from reality, and where experimentation is fast and very low cost, it will do at least as well as humans, but faster. That's a LOT of things, especially if it get access to a physics sim or any other "correctness oracle".
Published at
2023-04-16 00:29:23Event JSON
{
"id": "70294f423f081b9436f18230cb21234697b34df57d4dd5abbe5394200edd3076",
"pubkey": "c6de9f4954eddf9fd75170779653e2cb514aed692617a4d544864d84954da5d2",
"created_at": 1681604963,
"kind": 1,
"tags": [],
"content": "Tasks where a LLM can get immediate feedback from reality, and where experimentation is fast and very low cost, it will do at least as well as humans, but faster. That's a LOT of things, especially if it get access to a physics sim or any other \"correctness oracle\".",
"sig": "c456dd40be69e511465720eceecf81373dae64ddf6080ebf8b18c1be0b3df8676ced4d152b12cc6dd333c3d21a371fa54e294b6ae9e60f0abf9f9159eb6e0c38"
}
