TrustOps: A risk-based AI engineering process

We live in an era of artificial intelligence (AI) technology, which enables traditional HW/SW systems to help humans make wise decisions and even operate in intelligent ways. However, due to its inductive behaviors, this data-driven technology creates problems differ from those of conventional deductive systems. Therefore, the more AI is used in a wider range of industries and societies, the more problems and issues unseen before arise in human society. That's why most international organizations and nations released principles, and are struggling to find a technical methodology for assuring trustworthiness of AI. In many studies and papers, technical requirements are being presented individually and risk management is raised as the one of those. This paper intends to present an integrated and systematic engineering process that implements technical requirements in aspect of risk management.