Reading this article--while impressive--just shows us how far we are from a true AI. The definition of AI is "the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages".
In order to create a true AI, the machine itself would have to be as perfectly flawed as humans. It would have to possess the ability for irrational thinking, misinterpretation, forming a bias and then dismantling that bias. That would be true metalearning. We learn and interpret this world by making mistakes, recognizing these mistakes, then rectifying them. Additionally, in order for an AI to be fully integrated in our world AS a true AI and not a complex calculator, it would have to be able to recognize and factor in the numerous inconsistencies and exceptions in both our world and in humans. Creating an AI without these imperfections would make it nothing more than a machine, because as much science as philosophy provides on how and why and ways to recognize a human's thought process, there are always exceptions and information gaps.
As long as we are unsure of the exact way a human mind operates, a true AI will be nothing more than science fiction.
Let me be Xplicit; A google image search for "Xzibit machine learning" turned up (what I would classify as) underwhelming results. :(
edit: sorry, got drunkski last night