Here's the "podcast": https://www.youtube.com/watch?v=meu0CoYv3z8, incredible.
(Not hating on Rogan, just listen to any real clip and you'll see what I mean, e.g. https://www.youtube.com/watch?v=VvswhogSiY0)
EDIT: I listened longer and the AI actually is throwing in some "uhh"s and pauses. They don't sound completely naturally but it's still really impressive
i wonder if they have some extra input/coding to mark for inflections, etc
maybe they just tried a bunch of times and picked the best fit for the context