As far as AI, I suspect the limits of neural nets will become obvious and there will be an AI slump as investors pull back. R&D will continue, but often there's a lemming effect where if enough pull out at the same time, investors in general get spooked and follow. There is too much R&D money in it right now.

Neural nets will probably need to be combined with rule-bases, simulators, etc. to give AI more "common sense". Powerful pattern-matching alone is not enough, only part of the AI puzzle. There will be a slump until somebody figures out how to combine these well.