Overview of RAI Activities throughout the Product Life Cycle

• Stage 1. Intake
  • 1.1 Consider Previously Learned Lessons
  • 1.2 Determine Relevant Laws, Ethical Frameworks, and Policies
  • 1.3 Identify and Engage Stakeholders
  • 1.4 Concretize the Use Case for the AI
  • 1.5 Decide to Proceed to Ideation
• Stage 2. Ideation
  • 2.1 Define Requirements
  • 2.2 Identify Risks & Opportunities / Navigate Tradeoffs
  • 2.3 Weigh and Navigate Ethical Tradeoffs
  • 2.4 Write Ethical Statements of Concern
  • 2.5 Design to Reduce Ethical/Risk Burdens
  • 2.5 Accountability, Responsibility, & Access Flows and Governance
• Stage 3. Assessment
  • 3.1 Assess Requirements, Statements of Concern, Mitigations, and Metrics
  • 3.2 Exploratory Data Analysis
  • 3.3 Conduct AI Suitability Assessment
  • 3.4 Update Documentation
• Stage 4. Development/Acquisition
  • 4.1 Instrument AI to Promote Assurance
  • 4.2 Update Documentation
• Stage 5. TEVV
  • 5.1 Test System for Robustness and Resilience
  • 5.2 Revisit Documentation and Security
  • 5.3 Update Documentation
• Stage 6. Integration & Deployment
  • 6.1 Perform Operational Testing
  • 6.2 Train Users
  • 6.3 Establish Incident Response Procedures
  • 6.4 Revisit Documentation and Security / Roll-up into Dashboards
  • 6.5 Update Documentation
• Stage 7. Use
  • 7.1 Perform Continuous Monitoring of the System and its Use, Context, and Ecosystem
  • 7.2 Ensure Updating and Retraining
  • 7.3 Plan for System Retirement
  • 7.4 Record Lessons Learned