To me, it currently reads as if the path to recursive self-improvement is relatively clear. The only problem is alignment. Looking at the exponential increase, practically the explosion, of AI agents within OpenAI, it's clear that this growth will continue. It's still