Ah cool, I like this kind of thing.
It's also possible (and imo highly likely) that theism isn't itself hard-wired, but is a byproduct of certain tendencies/deficiencies in human thinking that *are* based in our neural hardwiring.
At 18:14 he says that 'Children will spontaneously invent the concept of god without adult intervention.' So perhaps 'hyperactive agency detection' always makes the leap to theism.
So, in order to maintain itself, atheism will need to do constant work correcting biases. In order to survive - at least before a 'change in fundamental human nature' (52:59), which he judges unlikely - it will need to be culturally transmitted successfully.
To prosper culturally, atheist societies will have to meet with comparatively more success at enabling human flourishing. Can they or are we better off indulging some, or all, of our biases?
Other points:
The vocabulary he uses suggests opposition - disabusing believers rather than informing them. Given that negative tactics (rejection rather than substitution) have generally failed (even the most coercive ones), this approach seems ill-conceived. That he uses overly oppositional language suggests possible bias - perhaps why he was unable to meet the 'religious belief is adaptive' implication proposed by the last questioner.
His ascription of intentionally manipulative, hierarchical structures to even primitive religion - that religion has been 'exploiting' biases thousands of years before science became aware of them - is odd. Presumably the first religious followers were just that:
followers, sharing experiences of the disembodied god agent produced by hyperactive agency detection bias.
The brain scanning argument is surely a non-sequitur: it is obviously not the case that because something has a correlate in neural activity it then has no metacognitive existence. When I look at a plate the parts of the brain that process visual information might light up. This doesn't mean that the plate has no independent existence.