Well, I never thought instant translation will be possible ten years ago. Google translate is doing literal magic nowadays.Think about it.
Some people in the conversation will be reasonable.
Others will be extreme.
And what is said will require an unbiased human to understand the context of the words in the entire conversation.
So they are going to have to spend time to listen to the entire conversation and analyse it.
So how can a machine automatically make a determination of whether things have gone too far?
And at what point do you cut off the conversation, given that this means censoring both reasonable opinions and also the disinformation?
Contextual speech recognition is not beyond reach. There are Grammar analyzing softwares that improve constantly as it works on thousands of articles.
And then there's the power of governments. How long do you think will it take once the government of a Global power decides that platforms like Clubhouse pose a threat?