IMO framing it as "get rid of the bots" is the wrong way to solve the problem because it's unsolvable, especially as we head more towards a world of content being generated by increasingly sophisticated ML algorithms.

It's probably best to embrace the bots and instead think about it as a problem of categorizing, quantifying, and applying more metadata to messages that let people, agents, or other bots determine the qualities about the information that they want to know.