This is very complicated feature to build efficiently and the rabbit hole is enormous. At some point you start considering machine learning, and the rabbit hole goes deeper and deeper.
I see the appeal of using AI to determine related content, this can be very handy on Support topics, which can almost be made “self service”. You post a question, ML picks up 10 candidates and just posts them as a reply you can either accept or delete. Similarly when anon hits something for support it is handy to show related content.
That said… gigantic enormous task. Not on our roadmap for next year. But @eviltrout is keen to experiment with AI/ML at some point and this is the type of project that could relate.
@codinghorror remains a massive fan of DJ random, cause he reckons it can outperform many many fancy algorithms (and it often does)