The GAMBLe framework provides a new analytical approach for optimizing AI-Driven Research Systems (ADRS), revealing that performance can vary significantly based on the interaction of components. This research indicates that careful selection of generators and assessors can lead to substantial improvements in performance and efficiency, challenging existing assumptions in the field.