Abstract: While building trustworthy AI has become a consensus worldwide, the rationality of trustworthy AI concept has always been a source of controversy. Critics challenge the concept by insisting that AI cannot satisfy trust conditions, while advocates attempt to classify AI or stratify trust to open up a space for trustworthy AI, but fail to provide a sound rationale. Restricted by the traditional interpersonal trust model centered on the trustor, critics and defenders are unable to conduct a comprehensive investigation from the dual perspective of human and AI, leading to critics making unilateral judgments and defenders providing failed defenses. Focusing on the trustworthiness and exploring the trustworthiness traits of AI would provide new perspective and feasible ideas to defense.