Visual recognition of the apple by AI only provides an input data; the soft grasping mechanism depends on separate capabilities.
The robotic system needs force sensors and a sophisticated motor control to translate this data into a successful physical action without crushing the object.
It's a necessary condition, but not a sufficient condition for delicate manipulation.
For example, a camera system can identify a water bottle, but without an appropriate grasping algorithm and articulated fingers, grasping remains impossible or destructive.
The direct causal link between visual recognition alone and successful manipulation is often overestimated; there are several critical intermediate steps.