Think of a play in a theatre, we say that there is a fourth wall that actors usually don’t “break”. It’s the same in movies, in general, nobody looks directly into the camera.
But on social networks it’s a bit different and it’s way more usual to see people looking and speaking directly to the camera. The main difference is, looking directly in the camera means speaking directly to your audience.
Looking out of it happens when the interviewee is answering somebody. The audience is then a third person watching the scene happening in front of them.
-> That is the main question, who are you/is your interviewee talking to?
It would be weird if for this video, I’d be talking next to the camera.
Think of a TV show for example, first the moderator looks directly to the camera, then to the guests.