The role of knowledge graph on responsible AI realization: research opportunities, gaps and challenges

As the development of AI algorithms the performance of AI systems has been largely improved. It, in turn, extends AI-empowered applications and makes AI play an increasingly important role in humans’ life. Thus, building Responsible AI which can be trusted by human has drawn much research attention in recent years. Linking various studies on Responsible AI through commonly accepted ethical values facilitates realizing it in the practice of designing, developing and deploying AI systems. Knowledge graph, as its advanced representation capability on facts and knowledge, can be used to represent and link heterogeneous information to support realizing Responsible AI. In this work, we reveal that knowledge graph brings opportunities for representing and processing unstructured information in legal documents, AI systems, and human interaction. Meanwhile, we intend to identify research gaps and challenges to smooth the future work.