Machine learning: 50 and high on the Gartner Hype Cycle

As I've recently written Gartner's Hype Cycle is a misleading hype itself. Which actually doesn't follow its own projected trajectory at all! It seems to be at a perpetual 'peak of inflated expectations'. As they say, old dogmas hardly die, the hype cycle is a point in case. It has no descriptive or predictive power and is thus misleading instead of enlightning. I predict that blockchain will be another technology that disproves the usability of the hype cycle (as I've written extensively before). But there are also technologies that outperform the hype cycle, the best current day example is in my opinion machine learning. This article shows that it has been at the 'peak of inflated expectations' for a number of years now, which is odd. But even stranger is that the cycle completely misses the fact that machine learning has by now proven itself beyond any doubt. It has outperformed even the wildest of expectations and is nowadays being used in almost every conceivable information system. There are of course limitations with regards to applicability and biases of algorithms that should be taken very seriously, but the number of useful applications is simply staggering. AI, and its sub-field of machine learning, is about 50 years old. It would be interesting to plot one of the most powerful technological trends of our era onto a hype cycle.