Six Steps to Responsible AI in the Federal Government: An overview and recommendations form the US experience

Abstract

This article discusses the challenges of implementing responsible AI in the federal government. It proposes six steps for operationalizing responsible AI, including concrete codes of conduct, tools to promote ethics and fight bias, evaluation benchmarks, and human oversight. The article emphasizes the need for tailored guidelines for different agency missions and the importance of user-centered design. It also highlights the role of organizational and mission-related factors in ensuring responsible AI.