I'm sometimes cynical about NNs too, but even if I were to take the hype cycle theory at face value, I have to admit it's possible that NNs are reaching maturity ("plateau of productivity") now, given that there were large hype cycles over them in the 1960s and 1980s, and given that they are starting to really work and are being deployed in large scale applications. But, this hype cycle theory isn't scientific, so all supporting evidence is anecdotal and subjective, right?
