My observation is that every wave of neural networks has resulted in a dead end. In my view, this is in large part caused by the (inevitable) brute force mathematical approach used and the fact that this can not map to any kind of mechanistic explanation of what the ANN is doing in a way that can facilitate intuition. Or as put in the article “Current AI systems have no internal structure that relates meaningfully to their functionality”. This is the most important thing. Maybe layers of indirection can fix

Source: Does current AI represent a dead end? | Hacker News