The risks from current AI capabilities that are diffusing throughout society are more pressing than potential future risks from Recursive Self-Improvement

Artificial intelligence research, products, and impact across society.

The process by which an AI system improves its own capabilities, enabling increasingly effective further improvements.