How coding agents are taking over the AI research loop, the AI-native routes to self-improvement, and what recursive self-improvement means for science, academia, safety, and the role of humans.
