>>9543044>An AI that can simulate conversation is still operating under an if/then/else model informed by extremely sophisticated programming.So I'm seeing two different possible interpretations of what you're saying here:
A) You mean this hypothetical AI* is literally just programmed to give specific canned answers e.g. "If input is 'How are you?' output is 'I'm doing alright'."
B) You mean this hypothetical AI is generating dynamic non-scripted output based on the input it receives, and you consider this an if/then/else model because their learning algorithm which allows them to generate their output is itself ultimately deterministic.
If the case is A, I don't think that's even really possible. And it certainly isn't something anyone would seriously try to take as an approach to AI because we already have much better case B type approaches that have long since been fleshed out available to work with today.
And if the case is B (which I think is the actual way this could work), then the sense in which that AI is deterministic isn't really different from the sense in which the human brain is deterministic. In both cases you can ultimately trace all output back to non-free cause and effect relationships, but this isn't the same as what's described in case A because the mechanism for producing output isn't itself just some "if A then B" formula, it's an approach to learning which allows for flexible dynamic output behavior.
*Which I'll point out is supposed to be so complex and well built that it can have a conversation with you where you ask random informal questions and get back coherent enough answers to where everyone who talks to it is under the impression it's just another human being producing said answers), hence why I don't think case A is even really possible.