Its use originated in game streaming, where the streamer addresses the audience as "chat" - a faceless entity that's comprised of thousands of observers that have meta awareness of the situation and steer the conversation by virtue of chipping in one short response at a time.
Chat is an emergent entity - a hive mind whose omnipresence hints at transhumanist takes on what efficient network communication might eventually evolve into.
The throughput can be astounding as well. While some chat groups consist of a dozen people, others count their nodes in the thousands. Just imagine a vast stream of voices, some relevant to the issue at hand, and others acting as outliers who poke around the margins, all self-policed by established rules of communication, rapport, and defining features such as visuals or "karma" (weight of reputation within the group).
The most interesting thing is that young people find it very easy, intuitive even, to communicate with this hypothetical 4th person entity. They are becoming adept at following many voices, as they speak in unison or overlap one another with different, often conflicting views, conversations, or even longer debates that go several layers back and froth.