I made the familiar point yesterday during a panel discussion that even if AI were approaching human intelligence, we're nowhere near human power efficiency (in "intelligence per watt")
But is that actually true anymore? We need huge power to train models, but once we have them trained, distilled and quantized, what do we have?
The human brain draws abt 20 Watts. An M4 doing local inference on a heavily quantized LLM won't draw more than 40 Watts.
Peter Bloem
npub1mj…aht0w
2025-11-28 15:06:56 UTC
Author Public Key
npub1mjj8dtcld0se03zd42wh9gqjg4mm04ya8azgssagnrvzlvkuq2pqmaht0wSeen on
wss://relay.momostr.pinkPublished at
2025-11-28 15:06:56 UTCEvent JSON
{
"id": "7634f69fd59314cff4c7d36c4c55ea167d578577766f0f2daf4307465a436378",
"pubkey": "dca476af1f6be197c44daa9d72a0124577b7d49d3f448843a898d82fb2dc0282",
"created_at": 1764342416,
"kind": 1,
"tags": [
[
"proxy",
"https://sigmoid.social/@pbloem/115627944575322273",
"web"
],
[
"proxy",
"https://sigmoid.social/users/pbloem/statuses/115627944575322273",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://sigmoid.social/users/pbloem/statuses/115627944575322273",
"pink.momostr"
],
[
"-"
]
],
"content": "I made the familiar point yesterday during a panel discussion that even if AI were approaching human intelligence, we're nowhere near human power efficiency (in \"intelligence per watt\")\n\nBut is that actually true anymore? We need huge power to train models, but once we have them trained, distilled and quantized, what do we have?\n\nThe human brain draws abt 20 Watts. An M4 doing local inference on a heavily quantized LLM won't draw more than 40 Watts.",
"sig": "e8c1f07bc22ca1edb09a518deaff89969339eec00bccf2a33657b0f52e307e66d354ddbb4f485e825d6dc62ed78b22a834610a9a967a4c974d648b7e5024251d"
}