At present, AbuseFilters, templates, etc. afford volunteers relatively deterministic ways of identifying content within Wikipedia, using tools like regex.
Per what @Soda and I discussed during WikiConNA (2025), we think volunteers could become more effective at identifying content that would benefit from their attention were they be able to identify said content using fuzzier, more probabilistic, methods. This, hypothesized, could be immediately useful in the context of community-authored Edit Suggestions.