OpenAI to Beat Pros at TI8?

For the future progression of AI technology, it is necessary to learn this game at a superhuman level. DotA 2 reaches the complexity level of real games like chess. Speaking of chess, in the 1990’s IBM developed a software known as Go that beat some of the best chess players in 2016.

DotA 2 was developed by Valve and is among the best titles of the eSports industry. The competitive version of the game is played by 10 players split into two teams of 5 players each. The players spawn in their own base. They play the game by exploring the map and taking objectives. Each player picks one hero from a pool of over 110 heroes. Each hero has unique items and abilities which enables them to win battles. The game is a battle between two teams and the mission is to destroy a megastructure known as the “Ancient”.

Over one million active gamers play this game. It arguably has laid down the foundation of the eSports industry. In 2017, the biggest competition of DotA 2, The International, had a prize pool of around $25 million. This is the biggest prize in the history of any competitive eSports discipline.

#AI bots just beat humans at the video game Dota 2. That’s a big deal, because their victory required teamwork and collaboration – a huge milestone in advancing artificial intelligence. https://t.co/UqIUhh9xFc

— Bill Gates (@BillGates) June 26, 2018

The learning algorithm of the Open AI Software was trial and error. The bots fought against one another. In psychological terms, it is known as reinforcement learning. It is also used in teaching children. The Go-playing AI developed by DeepMind also used this learning technique. In the initial stages, the bots make random movements across the map. They learn by a series of rewarding techniques to learn the playing environment. The reinforcement learning is quite effective in every training procedure to induce a sense of winning and losing.

The reinforcement learning techniques are quite simple and efficient and are used in training children and animals. So, it is logical that we will see its implementation in AI and machine learning. The bots were trained in a dedicated environment for extended periods of time.

Now the AI Program is all set to compete against the best players in the game at The International. The match will be held in August at Rogers Arena, Vancouver. With these simple yet effective techniques, we can hope that the AI beats the pros. It has done so before in the 1v1 matchup, but it might be a very different competition when playing with five compatriots. So, we might see the pros prevailing. We can only wait for the questions to be answered in August.