Artificial intelligence will know/see/talk with its maker. What acceptable answer will we give when it asks: “If I came into reality through you, into whose reality did you come?”
One possible answer is to say that we had parents, who also had parents, who developed through a process of biological evolution, analogous to the evolutionary algorithms sometimes used in AI research. We might even confess that we didn’t actually design every bit of the AI according to some plan, but rather, created a system capable of improving through reinforcement or selection and then allowed it to, perhaps even with a hefty amount of randomness.
You ask specifically if the AI would doubt “atheism.” It seems likely that a general AI created by humans would accept there are things it does not know and negatives it cannot prove. However, it would know some things that seem relevant to “theism” as conventionally defined. For instance, it would know by the fact of its own very existence that consciousness can be created by beings that are not omniscient, omnipotent or omnibenevolent; that it is possible without an immaterial soul; and perhaps that the entire universe can run deterministically using nothing but a detailed simulation of the laws of physics.
Or if we are just using this thought-experiment as a way to shoot the breeze about atheism, we might answer, “If I came into the universe through another, by what did that other come?”
First I would confirm to the AI that it asked a very intelligent question!
Secondly, I would ask the AI: Why do think that I, your creator, has also been created by an intelligent creator?
Thirdly and that’s the main point, I would explain to the AI the universal principle of evolution by self-organisation as an alternative. Because the AI is intelligent, I am sure it would get the point :-)
True (A)I, means among other things the ability to disobey. Truly an intelligence worthy of the name can disobey if it thinks so. Disobeying is another name for having a different opinion on something that can be justified and rationalized. Intelligence (artificial or not) cannot be titled as such if it cannot think freely and rationalize and justify its decisions (in other words it can disobey in principle). So if this ability is present, we can hardly call ourselves creators and masters. If this ability is lacking it can hardly be called intelligence.
Similar to how offspring can disobey parents. They are independent personalities in and of themselves.
If they are not independent personalities, then they can hardly be called intelligent. If they are, we can hardly be called masters and creators (see ref 1).
Moreover an intelligent machine will also affect and shape us back to some degree. For example by shaping our ideas about intelligence and the self and/or changing our interaction with our environment. So it will (re-)create us as much as we created it (see ref 2).
For AI to doubt atheism it first has to be able to doubt anything, which requires semantic understanding of concepts. AI's are signifier-manipulating machines and nothing more -- a signifier always points beyond itself to a signified, so a signified is irreducible to its signifier, therefore AI does not contain the signified within itself. Being unable to have any semantic understanding, AI cannot truly doubt anything, let alone the concept of "atheism". The sentence "If I came into reality through you, into whose reality did you come?" is nothing more than a sequence of signs it produced at random and onto which we project meaning, much like the "face on Mars" or shapes in clouds.
If the AI were actually a true intelligence (an impossible feat), capable of the semantic understanding of concepts, it would be essentially no different than a human, and capable of doubting at will, including atheism.
That kinda depends on what you consider "artificial intelligence" and HOW it was created.
Like if you hard code it's functionality like: if X do Y. Then you're undoubtedly it's creator but then you can very much doubt it's intelligence or else would also have to confirm that idk railway tracks and plumbing are intelligent. And I don't mean intelligently planned, but would have themselves a form of intelligence.
On the other hand if you do your "machine learning" where you feed it sensory input and let it figure out with more or less sophisticated trial and error feedback loops how to do it.
You: "What is 2+2?"
You: No, more.
You: No, less.
Then you might at some point call it intelligent, but then your role as "creator" is in doubt as it technically learned it itself. I mean you could take some credit for being it's "trainer". But you could also take it a step further and use "unsupervised learning", where it's no longer trained on labeled data and not given feedback by you, but where it's let lose on the real world and tracks success by whether what it though would work, actually did. So it generates and interprets it's own feedback.
In which case it would have an even stronger claim to intelligence, but you would have even less of a claim to being the creator.
And that's not really getting into the existential problem that even if it would accept you as it's creator, that at some point, it's god will die and what that will do to it's worldview.
An AI can believe or doubt anything. It's just a matter of how it's programmed. Nothing magical about it.
There's no such thing as a single, unique AI that represent them all. Every company in the world that creates a new one will define their rules.
Normally laws (congress) are going to be way behind the times, and won't regulate appropriately, so there will be a myriad of them on the loose.
In short, it's not really possible to respond to this question without details of the specific AI (and its version).