The case for using multiple people's AI models for value alignment

In a reply thread, a user argues that the query returned after sending a prompt to different people's language models represents the “sum total” of those people's values and mental models. They call this a strategy for value alignment, then offer a self-described “big if true” claim that the setup solves most of AI's biggest problems. They say work toward it is moving too slowly and invite questions to improve the ideas.