I've been attracted by this idea of perfection. What is the perfect way to solve this problem? Or the most efficient way to do this task? We humans are, unfortunately, limited in our analytical capacity. Our brains have evolved to make a monkey survive in nature, not play complex strategy games.

The brilliant aspect of algorithms like AlphaGo is that they only understand the rules of the game and are not limited or influenced by humanity's approach thus far (while this is not entirely true of AlphaGo, which did learn from top go players, an improved version called AlphaZero learnt go only through self-play and was superior to AlphaGo). That means they come up with brilliant approaches, that up till then, no human had even imagined. That's why game-playing AIs blow our minds, we are able to see a completely different yet effective approach to the game. We get a little closer to perfection.

AlphaGo is a wonderful documentary that triggered my interest in ML. A gripping, human story of how AI is developed and how it fares when pitted against a human. Was I the only one cheering for AlphaGo the entire time?