> I didn't get the reasoning behind the text-only examples at all. So the researchers can quantify how much of people's ablitity to detect deep fakes is coming

> I didn't get the reasoning behind the text-only examples at all. So the researchers can quantify how much of people's ablitity to detect deep fakes is coming from the video/audio and not the textual content and knowledge of politician's actual political position.