Investor Describes AI Refining the Concept of Good
Tweet questions the outcome of iteratively prompting models on moral definitions.
TLDR
@alth0u posted that repeated requests for an AI model to define or enhance the idea of Good result in ongoing improvements to the response. The author observes that this process eventually leaves one pondering the true nature of Good, suspecting the refined version falls short. The account presents itself as a general partner at a16z and comments on AI models, RL, alignment, and VC.
Combined views
597
1 Source, first seen 29d ago