Shmi comments on Agency engineering: is AI-alignment “to human intent” enough?