Comment by DrSiemer

We'll get to the point where we can mass moderate core knowledge eventually. We may need to hand out extra weight for verified experts and some kind of most-votes-win type logic (perhaps even comments?), but live training data updates will be a massive evolution for language models.