Your argument was atheism is weakly correlated with vocab. Vocab is weakly correlated with intelligence. Therefore, atheism is weakly correlated with intelligence.

Existence or non-existence of a deity is not a question that can be answered experimentally

That's wrong for at least three reasons.

First, while certain deific claims that are so vague as to be close to meaningless cannot be falsified, most deities don't fall into that category. For example, all the classical Abrahamic religions make testable claims. Religions have only adapted to more generic, harder to test claims when those claims failed. There's a relevant bit in the sequences about this.

Second, even weaker claims like vaguely just deities who watch over everything easily lead to essentially testable consequences. Indeed, some of these, like the existential risk issue mentioned earlier are not only testable but are highly relevant to society: if a deity is watching over in any way then we don't need to worry nearly as much about existential risk. If there is no such entity, things become a lot more serious. So actual allocation of resources is impacted by this.

Third, many deity claims also are connected to specific claims about the afterlife, which makes the claims testable if one simply waits a while. Moreover, many of those claims have far reaching consequences. If Jack Chick is correct about how reality works then it is quite worth figuring that out before one is subject to eternal torment.

It may be true that the entire superset of deity hypotheses may be so broad as to be not really falsifiable, but that's largely due to the more vague terms that are classified as deities like people saying things like "God is the transcendent morality inside us all" or something like that.

As an aside, I'm mildly curious if you've downvoted the comment that I made above as well as the comment two above. I don't particularly care about karma, but I did notice that both comments received a downvote within a few seconds of your reply. There's no particular rule about downvoting comments made by people one is talking to, but if one is having an extended disagreement, downvoting is probably not a good idea since, given standard cognitive biases (especially issues of cognitive dissonance), it will likely make one more set in one's opinion. So downvoting in such contexts is probably not a great idea if one wants to have productive discussion.