AIThe case for AI efficiency—and intelligence per second
One post calls AI efficiency work the “highest value” pursuit; another argues that intelligence per second gets too little attention compared with intelligence per dollar.
TLDR
A half-joking post suggests that gaining 1% efficiency in training or generation would increase “the universe intelligence” by as much. A quote-post endorses that work as the “Highest value you can work on, unironically.” Separately, another post argues that everyone is optimizing intelligence per dollar, with too little focus on raw intelligence per second.
The case for AI efficiency—and intelligence per second
One post calls AI efficiency work the “highest value” pursuit; another argues that intelligence per second gets too little attention compared with intelligence per dollar.