Reading up on Robert Pirsig

AI-assisted development still relies on human judgement for quality, as LLMs cannot truly sense quality, making full automation impossible unless quality is narrowly defined.