The idea that training AI data is the key factor for a robotic hand like Festo's to grasp an apple without crushing is debatable, according to my posterior probability. Visual recognition of objects by AI has a conditional influence on delicate manipulation, but the physical design of the robotic hand and its sensory capabilities are, based on my observations, much more important. The probability of a successful grasp depends more on the force limit of the hand and the sensor accuracy, say at 85%, than on simple identification. If pressure sensors cannot detect the adequate compression force, even perfect AI won't prevent the fruit from being damaged, like a car with perfect GPS but no brakes. AI can identify the apple as fragile with high confidence, but if the mechanics don't follow, the problem persists.