REAL Rating Framework Discloses Human-AI Collaboration Levels

Summary

The paper proposes the Reported Engagement with AI Level (REAL) Rating, a framework for disclosing how much generative AI is involved in products and the experiences surrounding them. It argues that existing provenance and disclosure practices often reduce the distinction to entirely human-authored versus AI-generated content, or rely on technical watermarking that is not clear to users. The authors say these approaches overlook the different risks and outcomes associated with different forms of AI use. REAL introduces six levels, numbered 0 through 5, to describe AI involvement in both product creation and user experience. The paper presents the framework’s structure and explains how it can be applied to text, image, audio, video, and software products.