MARIA: A Framework for Marginal Risk Assessment without Ground Truth in AI Systems
This addresses the challenge for software teams in deploying AI systems when ground truth is unavailable, offering a practical solution for risk assessment.
The paper tackles the problem of evaluating AI systems for deployment without ground truth by proposing a marginal risk assessment framework that compares systems to identify improvements and risks, providing actionable guidance for responsible adoption.
Before deploying an AI system to replace an existing process, it must be compared with the incumbent to ensure improvement without added risk. Traditional evaluation relies on ground truth for both systems, but this is often unavailable due to delayed or unknowable outcomes, high costs, or incomplete data, especially for long-standing systems deemed safe by convention. The more practical solution is not to compute absolute risk but the difference between systems. We therefore propose a marginal risk assessment framework, that avoids dependence on ground truth or absolute risk. It emphasizes three kinds of relative evaluation methodology, including predictability, capability and interaction dominance. By shifting focus from absolute to relative evaluation, our approach equips software teams with actionable guidance: identifying where AI enhances outcomes, where it introduces new risks, and how to adopt such systems responsibly.