<>
Microsoft is formalizing its approach to artificial intelligence development by aligning its internal safety frameworks with the NIST AI Risk Management Framework (AI RMF 1.0). Sarah Bird, Microsoft’s Chief Product Officer for Responsible AI, emphasizes that irresponsible AI outcomes often stem from experimentation conducted without a rigorous assessment of potential societal impact.
Aligning with the NIST AI Risk Management Framework
Addressing the Risks of Unchecked Experimentation
A primary challenge in the current AI landscape is the rapid pace of experimentation. Sarah Bird identifies that many safety failures occur when teams push technical boundaries without establishing clear boundaries for human impact.
Microsoft’s strategy to counter this involves embedding "Responsible AI" checkpoints directly into the engineering workflow.
Reducing Escalation Through Human-AI Workflow Design
Beyond technical guardrails, Microsoft is investing in research focused on human-AI interaction design. The goal is to create workflows that reduce the need for unnecessary human intervention while maintaining high oversight levels.
This research into "thoughtful workflow design" explores how to keep humans in the loop for high-stakes decisions while automating the routine tasks where AI excels.
Key Principles for Responsible AI Development
To maintain alignment with industry standards and internal safety goals, Microsoft’s approach relies on several priorities:
Related reading