Sarah tries to know the 250 or so people pushing the frontier of AI research. I asked her what's changed in how that group is thinking in the last 6 months: "I don't think every researcher doing frontier work at these labs feels like they're essential to the machine. With recursive self-improvement of AI research models that can improve the models themselves, we are a 1-2 years away from some sort of exponential intelligence. That belief is new within the last 12 months for a lot of…