For doing an interview over Skype or some other conference system, where you want audio without feedback of the originator spoiling the audio. That way you can have each participant only sending their voice, and not a mangled delayed set of the voices of all the others as well. Also allows a little local loopback so you keep your voice level consistent, otherwise without sidetone and with headphones on you tend to shout after a while as the brain is getting no audio volume feedback.
If you are using a good mixing console this is done for you automatically, and the best ones do it for each participant in the system as well so all audio is as good as possible. Good audio there, though the bald presenter does need to turn down the high end a little or use a de-essing filter on the microphone as well.