The central question if you notice is that - these complex AI programs are designed to perform one specific task and be good at it to the extent that they can defeat the best human in that task. A normal human may not be the best in every task that he/she does, but overall they do multiple task quite good from making breakfast, driving to work, spending time with family/friends etc. So there are different aspects of empathy, evolution, emotions and combination of all of these things make us the great humans (Ensemble Learning). However, AI is not designed with these ideas in mind. Humans are very good in learning knowledge in unsupervised manner and transferring that knowledge to learn new things - (Transfer Learning).

Even though it can't be widely generalized, it's still a very big deal. To put Google's AlphaGo win in perspective, let's consider the other two major victories of machines over humans in recent com... This is because AlphaGo displayed true cognition. Keep in mind we're not talking about Artificial General Intelligence (AGI) but it is Artificial Special Intelligence (ASI). Fairly significant, but not nearly as significant as presented in the press.