The esports industry has grown and diversified itself enough to guarantee its own long-term survival. Electronic sports like Mobile Legends: Bang Bang are played by millions around the globe and their professional tournaments attract increasingly larger crowds. Many also enjoy betting on MLBB and other titles using the online services of top bookmakers. But despite the growth and the fact that great companies are supporting their development, creating intelligent AI bots for these types of games has proven to be a daunting challenge that required outside companies to step in.
The Complexity of Esports
Just like survival equipment designed for mountaineers is extremely sophisticated compared to your casual one, electronic sports tend to be a lot more complex than the average video game. And that’s because they’ve been designed for people who love to compete and always search for a novel strategy that can win matches. Millions of dollars are on the line in professional tournaments and the audience wants to see a great show. Therefore, esports need to have a lot of depth, so that they allow players to be as creative as they want.
Multiplayer Online Battle Arena games constitute a very important genre in the esports industry. And their complexity is phenomenal. Such games are made up of around 800 – 1000 small components that you need to learn before you can understand them at the most basic level. Forget about mastery. You need to play League of Legends for 6 – 12 months just to learn how it works!
The reason why MOBA games have so many elements is simple. These games feature RPG and RTS elements. When you have a lot of unique heroes with abilities, who can buy items to become even stronger, the game you’ve created becomes not only unpredictable but also hard to learn.
MOBA games tend to have 100+ heroes and each of them has at least 4 abilities. On top of that, you get to choose between a few hundred items that you can buy with gold. That alone creates 600 – 800 elements that need to be learned. And apart from that, there are many other elements and mechanics that are part of the game.
The Dynamic Aspect of Esports
To keep their games fresh and exciting, companies like Valve, Riot Games, and Moonton regularly add new content and make a wide variety of changes. In games like chess and Go, the rules never change. But in esports, the system changes significantly at least once or twice a year. And in some cases, every month! When that happens, players need to re-analyze the game and figure out what the new meta or the new ways of winning matches will be.
Strong heroes can be nerfed and weak heroes can be made really strong. In turn, these changes lead to novel strategies that nobody has thought about before. In every match, the battle takes place between 10 players and the format is 5v5. Even in 1v1s it would create a lot of unpredictability. In 5v5, the effect is exponentially greater.
Given the two problems discussed above, it’s no wonder why truly intelligent AI bots are hard to create for many of the world’s most famous esports. It took years to create strong chess bots and competitive video games are a lot more complex than chess. But despite not having the ability to solve the problem themselves, companies like Valve have teamed up with AI companies such as OpenAI and DeepMind. These companies had already had great success in creating powerful bots for board games like chess and Go. So they were expected to come up with something brilliant for esports as well.
One of the greatest success stories related to smart esports bots is OpenAI’s. The company introduced its little monster in 2017, during the Dota 2 World Championship, called The International. And it defeated the strongest players in the world. Then, in the next two years, the company gave the bot the ability to control an entire team of heroes and that team defeated OG, the strongest team in the world at the time.
The second success story is DeepMind’s. After it conquered Go and chess with AlphaGo and AlphaZero, DeepMind did the same thing in one of the most iconic real-time strategy games of all time, StarCraft 2. Its creation, AlphaStar, defeated with ease the greatest European and South Korean players in the game.
What the Future May Hold
If AI bots can quickly master highly sophisticated games like Dota 2 and StarCraft 2, imagine what else they can do. We are moving toward a future that’s a bit scary. But it could also be amazing. It’s up to us to put AI to good use and make sure it doesn’t get out of control.