I, an atheist, think that a computer system complex enough to learn cause/effect could make enough mistakes to falsely come to same types of mythical conclusions that led to religion.
Hopefully, they’ll iron out those bugs in the consequent release… 
If anyone’s interested, there’s actually a book (fiction) that cover’s this 
When H.A.R.L.I.E. was One by David Gerrold is about a computer that was built to “simulate” human though processes, and is actually sentient. In addition to some fascinating discussions about life, intelligence, communication and love, it decides to determine whether or not God exists. Harlie starts from Voltaire’s quote “If God did not exist it would be necessary to invent Him.” So he designs an infinitely scalable computer system, to be programmed by himself (since even a team of human minds has a ceiling as to how complex a program it can handle). This is the Graphic Omniscient Device–G.O.D.
There’s a bunch more with office politics, gravitational detection methods, and one or two good sex scenes. Definitely try to get your hands on it, for the general philosophical discussions between Harlie and his “mentor” Dave. And yes, that does lead to at least one use of “Harlie, David’s son” But no “I’m sorry, Dave, I can’t do that”, though.
The questions become important and difficult rather quickly when the topic is AI. How do we define it? How do we make it? How do we know if we’ve really accomplished these goals? As this article illustrates, the AI could include biological components. In the linked article scientists report on experiments in which they isolated rat neuron cultures and connected them to a robot “body”. If we were able to wire up a rat’s brain to control a human’s body (leader of the free world jokes aside) would this be a human or a rat? What if we were able to build an AI and install it in a human? What it we took a human mind and placed it in a machine? Tricky stuff. I’m of the opinion that the ultimate outcome will be a merger of biology and machine rather than a division. The result will be a new form of intelligence better joined then either ever was alone.
As for the “modal logic proof of God” fallacy, all that “proof” illustrates is that whatever is necessary for existence (label it NE - Necessary Existence) exists. Simple. Once you start assigning qualities to NE (such as God, Yaweh, Merciful, Good, Evil, Carrot Top fan) we now require a leap of faith. It is this point that the religious defenders of the proof tend to gloss over (at least here on this board - I have seen this objection raised several times and not one person has been able to counter it with anything stronger than simply ignoring this glaring weakness). You see, NE could be quantum mechanics, string theory, a giant pixie fairy, a mindless brute, a complex awareness, or none of the above. To call such a thing “God” or “Him” is an attempt to take a blank canvas and paint it with one’s own bias.
DaLovin’ Dj
It could be that the most rational position for an AI to take would be agnosticism - coupled with the scientific method, the poor thing might conceive the notion that it is necessary to find proof of the existence of God,
rather than being satisfied by ‘mere’ faith in an unknowable being.
To obtain such proof it might be necessary to destroy the entire universe in a Tiplerian collapse…
like I said elsewhere, creating artificial intelligence could be a risky business.
Sci-fi worldbuilding at
http://www.orionsarm.com/main.html