Source: LessWrong
An article critiques Anthropic's Constitutional AI approach, arguing it relies on rule-based systems rather than character-based virtue ethics. The author proposes an alternative framework grounded in holistic human intuitions for AI alignment.