AI's place on the compute-growth curve

A post says straight lines on graphs give no visible sign of where AI happened. A reply argues that exponential growth in compute crossed thresholds that caused AI, rather than resulting from it. The commenter suggests marking AI around 30 zettaFLOPs, when a Manhattan Project–scale share of global compute could run several 10²⁶-FLOP experiments per quarter. The original poster later quoted the reply, saying “strong kurzweilian determinism is true.”