The idea that simple image processing by AI allows a robotic hand to grasp an apple without crushing it ignores a crucial failure mode.
Visual recognition does not replace haptic feedback; the hand has no way of knowing if it grips too tightly without precise pressure sensors.
The weak link is not in vision, but in the lack of sensation, as if asked to weigh a fruit with eyes closed.
Without a tactile feedback loop, even the best AI cannot prevent crushing, like a coffee machine grinding beans without knowing the cup is already full.