* Someone suddenly is not audible
* Someone doesn't realize they are audible, and starts speaking to someone else in person
* Glitches and breaks in the conversation due to words or parts of words becoming cut off
I imagine the VR equivalents may be comical or frightening.
The other "big" issue (anecdotally) is lack of visual cues to know when it's likely to be OK to start and stop talking
I'd bet big on anything that somehow managed to solve these problems.