The really interesting question is why is the human better than the autopilot for strong cross-winds?
Generally most pilots would agree that automated flight control is more precise than manual control. Automated height keeping, heading keeping and (normal) landing are all as good as a pilot at the top of their game.
It is, after all, a simple exercise in control law. Simple mathematical rules of input and response which automatic systems can be tuned to do extremely well. In the human of course the mathematics is replaced by "learned response" also known as "experience" and "practice", but it is still basic control law in action.
So why does the automatic system go wrong when it gets a bit gusty? Reaction times for automatics are well fast enough for wind shear and gusts. So there is a strong implication that the human has access to some subtle extra data that the automatics just do not have. But what?