The fundamental problem with this article is that it restricts it's analysis to a model composed of a single neural network. Any "general intelligence" will und

The fundamental problem with this article is that it restricts it's analysis to a model composed of a single neural network. Any "general intelligence" will undoubtedly come from a complex of neural networks, much like the human brain. I would argue on that front we are closer than the article implies. Look no further than the derivatives of AlphaGo. Playing such games is a form of generalizable reasoning.