I think that part of the issue with modern AI is finding an actual application. Most of it seems to be used to replace jobs which shouldn't exist (writing rage-bait "journalism" online) or have needless overhead (using AI to convert points into reports, then using another AI to summarize the report back into points).
Modern neural networks are very good at finding patterns in data, but aren't very good at making sense of the patterns (part of why they can generate human-like content with correct grammar and concrete facts which is totally nonsensical due to contextual inconsistency - like how the same AI can tell you about what Elon Musk has done lately while insisting he died in 2018).
Hard to know exactly where that will end up going over the next decade but finding patterns in large data has its applications (some being good - like drawing attention to patterns in scientific observations - and some being nightmarish - like tracking people).
The question I always have is what humans are actually trying to accomplish (sure, we won't agree, but I think it is a disagreement worth keeping front of mind).