The assertion that training AI data is the main reason a Festo hand can grasp an apple without crushing it overlooks a factor with a much higher influence score. True success is around the 70th percentile for design, including pressure sensors and mechanisms, compared to about the 30th percentile for AI. For example, if the robotic hand is not built with materials and mechanics allowing delicacy, no amount of image learning will prevent it from crushing the apple; it would be a zero coefficient on the goal. AI adds recognition accuracy, but mechanical capacity is the prerequisite, the common denominator of the entire process.