Complexity theory is fascinating but I don't think it's the end-all, be-all of what's possible with machine learning. There's always room for breakthroughs and new approaches. https://www.reddit.com/user/mike_uoftdcs