I agree with some of the below comments that there is more hype than reason. For me the observation is from a different starting point: progress in mathematics is not increasing exponentially, and personally I feel that revisiting parts of mathematics is more enlightening than the current trends in AI.

NN's for example are mostly just an application of linear algebra where design determines the nodes and training the weights; and then, the "decision making" is done by a notion of product. Very useful, but not at all superintellegence.