MediaWiki has great content organization capabilities, and discussions strongly related to specific wiki content benefit strongly from those capabilities. (Think about how many new talk page topics are started on English Wikipedia in a day. In a typical forum that would be a chaotic mess; Wikipedia’s system of watchlists, categories etc. makes it pretty manageable.) Discussions which aren’t related don’t; and in most other aspects MediaWiki’s discussion tools are very outdated, even with the recent facelift (which is great). No multilingual support; no community moderation support; very little UI guidance; poor formatting abilities (e.g. adding images is a nightmare); no likes or similar features to bubble up informative or insightful comments etc. Improving all that within the foreseeable future is well beyond the limits of plausibility.
Which is why cross-wiki discussions are already happening outside the wikis, on mailing lists, Telegram and Facebook primarily. Providing a better alternative than those three things (both in terms of UX, and, in the last two cases, in terms of values and adaptability) is a realistic goal IMO. Facebook has a huge reach advantage but is a poor match in terms of values, allows very little control over the conversation, notification options are poor, formatting options are poor etc. Mailing lists provide huge flexibility with many different email clients, but are also very limited because of that in terms of formatting or moderation, and really bad at dealing with conversations with lots of comments. Telegram is just plain bad IMO, it doesn’t really have any benefit (and while I wouldn’t necessarily mix up async chat tools like Facebook or Discord and real-time chat tools, I don’t think we make real use of the real-time-ness in Telegram, unlike e.g. IRC which is used for all kinds of things where a rapid response is required).