Dawkins uses an interesting analogy comparing religious instincts to a moth's attraction to a candle flame.
Dawkins describes religion as an accidental byproduct of a whole range of psychological predispositions such as a child's instinct to believe and obey its parents - he might have mentioned hyperactive agency detection as well.
This is almost certainly true but is religion more than that? Dawkins declines to speculate on whether the byproduct - religion - is an evolutionary advantage, only that the instincts which resulted in religion were advantageous. In this regard he is at odds with Jonathan Haidt who describes how it may still serve a useful function...
In other words, Dawkins and Haidt agree on how religion got started, but Haidt goes further and insists that religion itself was an advantage to humans as a social species. To go back to Dawkin's analogy, the instincts that cause a moth to kill itself in a candle flame were advantageous to its ancestors, but the byproduct - death by candle flame - is not. In this regard Haidt would reject the analogy.
Significantly Dawkins finishes his answer in the video with the comment that "the religious byproduct is either neutral or ... well I, we don't even need to say whether it has an advantage, it doesn't matter..." Was Dawkins going to say that religion is either neutral or harmful? I don't know but he stopped himself from finishing that thought.
I tend to disagree with Dawkins. I think Haidt may be onto something.