Trustworthy artificial intelligence

Abstract This paper develops an account of trustworthy AI. Its central idea is that whether AIs are a matter they live up to their function-based obligations. We argue this serves advance the literature in couple important ways. First, it provide rationale for why range properties widely assumed scientific literature, as well policy, be required AI, such safety, justice, and explainability, (often) instantiated by Second, we connect discussion on AI industry, sciences with philosophical trustworthiness. extant accounts trustworthiness philosophy cannot make proper sense our compares favourably its competitors front.