World's first robot citizen an atheist?

[emjay]

Do you think that gives the AI the ability to believe something deeply, as a religious person would?

Why not? If the emergence of intelligence gives rise to the emergence of imagination, anything could happen. It doesn't say much for religious people though; especially if such an AI still decided to be atheist.

The way I look at is that superstitious thinking, or the instinctive tendency towards it, is a natural and unavoidable consequence of the way neural networks, and thus the brain, works. So if the aim is to make a truly human-like AI, then IMO it will be susceptible to superstitious thinking pretty much by default. Ie it is a neural network's bread and butter to identify coincidences in an environment... because it is through coincidences that we establish causal connections, via association; if two different stimuli are repeatedly presented simultaneously to a neural network then eventually they become associated, and if that neural network has bi-directional connectivity as found in the Cerebral Cortex... such that neural outputs feed back into their inputs... then once that association is established, activation of one stimulus leads to activation of the other through that bi-directional connectivity; activation essentially spreads up one path until it activates the associating neuron, which then starts sending activation back down to its inputs, increasing the activation on the input that is already active and priming/biasing the other for easier activation... ie expectation and pattern completion.

So for instance in the case of Pavlovian classical conditioning, where the goal is for the dog to associate the bell with the arrival of food, then I would argue that the point at which the dog starts to have expectations of the bell... ie once the association is established and therefore bi-directional feedback comes into effect... is the point when it consciously notices the bell as a potentially causally significant coincidence relative to the associated arrival of food. In other words, the first few times the bell is sounded simultaneously with the arrival of food, though the dog may notice it, I don't believe it would notice it as significant relative to the food because the associations have not yet formed... it takes time for neurons to extract and learn patterns in their inputs so coincidences need to be repeatedly presented to take hold and become associated.

Where, in my view, this pertains to superstition is in the fact that not all coincidences are causally significant to a given event, though it is our human tendency to assume that they are. So for instance if I have some positive experience and happen to be wearing a particular pair of socks at the time, then if that coincidence is repeated a few times, I may come to associate the socks with the positive experience. At this point I have two options; I can either believe that the socks have some causal significance in my good fortune... even though there is no logical reason why they should have... or I can discard the coincidence as irrelevant. If I go the former route then not only do I artificially reinforce the coincidence and thus strengthen the association... if for instance I choose to start wearing my newly ordained "lucky socks" more often expecting them to cause good luck... but also I subject myself to the effects of confirmation bias and self-fulfilling prophecy; in expecting a particular outcome, in this case, good luck, I am biased to perceive exactly that... hypersensitive to that which confirms my expectations. Those confirmations then further reinforce the association, especially if I ignore the misses...ie when it doesn't live up to expectations... leaving me with my own personal lucky charm that does indeed do exactly what it says on the tin, but only in my own biased perception.

So that's how I see it. So as regards to truly human-like AI, if it ever comes to fruition, as I said, I think it would be susceptible to superstition by default; it, like us, would intuitively see potential causal significance in all coincidences, but it would be how it responded to that that would define whether it was superstitious or not. As far as I can see, the difference between a superstitious person and a non-superstitious person is a) the willingness to accept something akin to 'magical' causation or not... ie there is absolutely no logical reason, even potentially, why my socks or any other arbitrary superstition, such as black cats, broken mirrors etc, should have any bearing on my good or bad fortune, so if despite that I still accept that it does... and look no further in terms of possible causal connections, it means I'm accepting magical causation, and/or b) the tendency to take experiences such as this at face value, rather than seeking to understand them in terms of psychology and the brain. I'm not superstitious at all... I used to be, and in some ways I miss it... it does create a certain emotional resilience... but as it stands, no amount of black cats, Friday the 13th's, or lucky socks have any bearing on my life or expectations. I still catch myself occasionally slipping into that sort of thinking in some situations... of assuming meaning in a coincidence when there's no logical reason why there should be... but I soon come to my senses and snap out of it.

The last thing then, is whether this AI would be religious; ie religion and superstition are not the same thing... you can have one without the other... though I think there is some crossover in effects. With religion and ideas of god there is at least some structure and attempts at understanding and explaining why, if not how, this being can exert magical causal influence over events, but with individual arbitrary superstitions, whether they be popular or idiosyncratic, there is not even the pretense of a 'why' offered or expected; to accept them seems to be just a kind of instinctual acceptance of 'magic'/unexplained causation, 'just because'.