This article might appear to be rather lengthy but it's worth reading every single word. The author, Gary Marcus, provides a short history and comparison of deep learning, symbolic AI and a hybrid approach. Spoiler alert: One size doesn't fit all, so he recommends the hybrid model.
A few paragraphs into the article I felt deflated, wondering if AI will always be one of those technologies that is stuck in place and constantly described as "a revolution that's just x years away", but x never gets here. The more I read, however, the more optimistic I became and began to appreciate the case he makes for the hybrid approach. In fact, the author has me so convinced that my next step is to go buy his book, Rebooting AI.