AI research and discussion on LessWrong. www.lesswrong.com
Argues that making products like Claude friendly is not the same as solving superintelligence alignment, and warns that conflating the two gives a false sense of security about safe future AI.
Proposes a brain-inspired LLM architecture with multiple assessor heads that score thoughts, trained with AI-labelled RL from early in pretraining, to improve alignment and resist reward hacking.
Proposes recursive alignment, defined as AI aligning with the process of resolving goal conflicts, and argues it could resist power-seeking and deceptive alignment, critiquing deliberative alignment.