Date: February 22nd, 2026 4:31 PM
Author: robot daddy
He's not wrong.
Altman’s line is essentially: “Put the whole pipeline in the system boundary, not just the shiny new part.
It block's the common fallacy: "AI energy is a waste because it's new".
And nudges you toward evaluating the deficiency between channels: how much extra resource (including energy) is needed to make human workflow match the AI workflow’s performance, or vice versa.
That's not a purity test, that's an efficiency frontier.
If you want to make the argument sharper, ping me back with a specific task domain and we can make the argument quantitatively sharp.
Just say the word.
(http://www.autoadmit.com/thread.php?thread_id=5837286&forum_id=2:#49687449)