If you have ever played videos games before then you are probably familiar with AI. Whether the game is PacMan, GTA, or Need for Speed, you will find an AI to interact with. However, given spending enough time of video games, most people can figure out that AI in these games are rather “dumb”. They have a pattern and they follow it religiously, if we do something, they do something, if we don’t do something, they do something. A perfect example would be how AI interact in certain games such as Call of Duty or Super Mario, they follow a Finite State Machine pattern such as follow
In order words, up until very recently, AI has been rather a set of conditional statements generated beforehand. In this case, translating the above pattern to something similar to Ruby where evade, health and wander are methods.
self.evade if self.health < 30
self.wander if self.health > 90
All these statements are rather simple and for the longest time, gamers were happy with it. However with evolved games come evolved gaming experience. Gamers demand a better, harder yet more enjoyable experience while gaming, come the era for better AI.
Games such as Civilization (an RTS game focusing on managing resources and diplomatic actions) uses a more complicated known as the “Monte Carlo Search Tree Algorithm:
This algorithm shows a much more dynamic relationship between a user’s action and the AI’s reaction. AI no longer follow a set of pre-set commands, but rather it uses random trials to solve a problem. In short AI in certain games now has an analytic mindset. Of course this is still not enough to impress lots of gamers. But what if i tell you, AI is about to reach the maximum point of being the best gamer of all games?
Chess is a game of strategy and resilient. There are a total of 10⁴⁵ moves in chest and to master and know all this moves take a genius in the human race. Enter 2 very iconic chess world champion: Garry Kasparov and Magnus Carlsen. The peak of humanity when it comes to chess, scoring 2851 and 2882 Elo (chess skill score)respectively. Both numbers when compared to DeepMind’s AlphaZero AI are mercilessly smashed, with an estimated Elo rating of 3600. While neither Magnus nor Garry have played against AlphaZero, most professional chess players would bet money on AlphaZero due to the fact that Garry himself lost to DeepBlue AI whose Elo was merely around 2700.
Moving forward to recent years where E-sports has become more and more popular. Dota2, a strategic online game with 115 playable characters and hundreds of item builds and customization within the games. The combinations of movement and actions in the game is far greater than 10⁴⁵ moves compared to chess, and to top it of, the game attracts its best players into its tournament where the prize pool last year in 2018 is: $25,532,177.
Introducing Elon Musk- sponsored AI, who made a surprise appearance to go against the top talents in the tournament. The match featured many funny and interesting moments to all players, especially this moment:
While most professional Dota matches last around 30–45 minutes, the match ended in 25 minutes with AI constantly stomping the Humans team. This moment became an iconic moment for the Dota community and become a tradition where every Human would rematch against this AI team every tournament. Spoiler alert: AI is still in the lead.
So with all the craziness going on in the gaming industry, one can’t help but ask what is going on with other industries where AI present. The answer is similar, AI has either triumph against human or slowly reaching the point of going above and beyond what our mind can process, eg: Algos trading in finance, automation in manufacture, automatic driving in transportation. While most people are scared about this, I am stoked to see what the future holds for us as AI’s capability improve and help our lives become easier.