Nope. They'll just outsource this to underpaid contractors like Facebook does.

Even better, if they could leverage LLMs, they could maybe automate a large part of the initial moderation. They have already acknowledged that these models were likely trained on their data. Why not feed new posts back into it and determine about relevance (on vs off topic), attitude (compassion vs hate) or quality (written by a bot)?

There's arguably more unpaid moderators on Facebook inside groups than reddit. FB claims 10million groups while reddit says 3.4 subreddits

reddit is an order or magnitude smaller so in actually surprised the group numbers are so close which only proves the communities is a core feature of reddit