Center for Global Development

In a new CGD blog, Tim Ohlenburg and Rikin Gandhi argue that evaluating AI interventions should go beyond model performance or even traditional impact evaluations. As AI tools, products, and users evolve, evaluation must become a continuous process of learning and improvement.

Drawing on Digital Green’s FarmerChat platform, the authors outline a four-level framework for assessing generative AI interventions—from model quality and user engagement to behavior change and real-world development outcomes.

Continuous Improvement Through Evaluation: Building Better and Better Generative AI Interventions How do you evaluate a development intervention based on generative AI, like a tutoring tool or a chatbot to provide advice to farmers? The Generative AI Evaluation Playbook lays out a four-level framework for evaluating generative AI interventions through a process of continuous assessment.