Data-driven AI is an approach in which systems learn patterns, relationships, and predictions from data rather than relying primarily on manually specified rules. It is well suited to dynamic environments where sufficient relevant data is available and the system must adapt to changing patterns.