Before I begin, let me put a disclaimer here: these are hastily assembled thoughts, engaging with the subject at hand before spending hours or days reading up on it, pondering it, and defining all my terms more effectively. If you like, consider this whole post a placeholder for later thought.
Personality, the will, sentience is solely a product of the breath of life, given by God, rendering wholly artificial intelligence impossible to create.
If you took a human person, replaced his limbs with prosthetics, compensated for the destruction of his nerves with some manner of electronic signals, gave him other replacements for his original organs or viscera, rigged up an elaborate support system for his brain: all this still would not make the intelligence itself, the person, artificial. It isn’t all mechanical. I disbelieve in a mechanical mind (though certain aspects of a generic human brain, to a certain extent, give physical or mechanical evidence of the processes occurring therein).
If a neural network produces anything, it does so through training. It doesn’t actually have its own intelligence to go on, just the promptings of an actual human (or, perhaps, a whole lot of data gleaned from a great many actual humans).
Can it, per se, ever recognize humor? Consider the InspiroBot Inspirational Images Generator. The generator generates; the human looks at the image and caption together, and that interaction is where the humor happens. It is humorous because the human mind recognizes the absurdity. It is humorous because of the human mind being struck by the unexpected.
I suppose you could argue that the generator creates the humor by presenting the unexpected. But I would then argue that the generator generates as it is trained to generate, making this ultimately a human creation.