Training the AI model for image recognition is necessary, but it is 75% likely that it is not sufficient for a Festo robotic hand to grasp an apple without crushing it. The mechanical capacity and sensors of the hand are at least as critical, if not more. For example, if the robotic hand does not have pressure sensors to detect the delicacy of an apple, AI cannot prevent crushing. The probability of a successful grasp is conditioned on 90% by the physical capacity of the hand, and only 10% by visual recognition alone.