@jeffreyhuber Good point: “For roughly three years the binding constraint on AI inside the firm was the model itself: was it smart enough to do useful work? That question is, I think, now settled, and not in a borderline way.”
http://x.com/i/article/2066314130827673600

