Improving verifiability in AI development

OpenAI Blog · 6 years ago

A consortium of 58 researchers across 30 organizations published a report outlining 10 mechanisms to improve verification of claims about AI system safety, security, fairness, and privacy. The report involves institutions including the Centre for the Future of Intelligence, Mila, and the Center for Security and Emerging Technology. The mechanisms provide tools for developers to demonstrate AI system properties and for policymakers, users, and civil society to evaluate AI development practices.