Standards for AI governance: International standards to enable global coordination in AI research & development

Abstract

Standards are an institution developed by societies to coordinate. For millennia, they have reduced barriers, increased efficiency, and protected consumers. Standards can be applied to Artificial Intelligence (AI) to support safe and beneficial development and use. One mechanism to develop AI standards is international standards bodies (ISBs), such as ISO and IEEE, which possess technical expertise, financial resources, timely information, and effective institutional knowledge. Yet AI researchers and organizations are not fully engaged with ongoing standardization activities. They can change this by engaging directly with standards committees, joining working groups, or participating in standardization efforts at other qualified organizations. Additionally, there are paths for researchers to pursue their standards initiatives that can later be brought to ISBs. Research is needed to identify essential standards desiderata and to identify effective strategies to make standards for AI governance a reality. Standards can serve as a tool to spread a culture of safety and responsibility in AI research and development. These standards can also ensure that safety features are built into the systems, promote best practices in the industry, and help support international agreements on AI. – AI-generated abstract.