OpenAI bot won the top three professionals in Dota 2

Danil "Dendi" Ishutin attacks the positions of the OpenAI bot at The International on August 11, 2017. Live Tournament Screenshot



Yesterday at the Dota 2 International Championship in Seattle, the bot created by the OpenAI organization defeated one of the world's best players in Dota 2, Danil Ishutin , famous professional circles under the nickname Dendi ($ 735,449 in prize money in his career). The 27-year-old Ukrainian surrendered in the middle of the second game.



According to the rules of the one-on-one match, a player was considered the winner in each game who would commit two murders or destroy the enemy's tower. In the first game, OpenAI dominated and won in less than 10 minutes, and Ishutin seemed amazed at the bot's capabilities. At the beginning of the second game, the bot made the kill, and soon Dendi stopped the game, acknowledging defeat. “This thing is scary,” said Dendi in front of a huge crowd of spectators. Ilon Mask exulted .



Thus, the OpenAI bot remained unbeatable in confronting the world's best players in Dota 2. Previously, he celebrated a victory over Arthur "Arteezy" Babayev (No. 1 in the overall rating) and Syed "Suma1L" Hasan (No. 1 in the 1v1 rating).



Dendi vs OpenAI bot





Dota 2 is a difficult game with hidden information, where players have to plan actions, attack, cheat and deceive the enemy. There is no obvious correlation between the player’s abilities and the number of actions per minute, although the bot has an indicator of the number of actions per minute approximately the same as in humans. Nevertheless, players note that the bot has gained an advantage due to faster response and extremely accurate movements, compared to a live person who clicks a mouse.





The staff of the staff of The International Championship raised his hand with a USB flash drive, on which the bot was recorded. Professional Dendi stands in the background, waiting for the start of the fight. Photo: OpenAI via YouTube



“What we have shown here is called the overall learning system,” explains Greg Brockman, co-founder and CTO of OpenAI. “She still has a number of limitations, but she is already able to defeat the best professionals in Dota.” This is a step towards building more general systems that can be trained in more complex, complicated and important real-world tasks, such as the surgeon's profession. ”



The OpenAI bot learned to play Dota 2 by conducting a large number of gaming sessions against itself. The training took two weeks. During this time, the path has been traversed from random random actions in the game to skills sufficient to defeat the best professionals. The developers did not put any strategies into the program, did not use the help of experts. The bot just started from scratch and played with itself, step by step making small improvements in the game until it reached the professional level.





However, in its present state it is unlikely that a bot will be able to compete in a big game, where teams of five players usually play. Still, one-on-one matches are a simplified version of Dota, but in team games there are much more different strategies and specific techniques. In an official blog, the OpenAI organization stated that creating a group of bots to play against a team of people is the next goal.



For OpenAI this is a definite achievement. This non-profit organization was founded in December 2015 by well-known entrepreneurs Ilon Musk and Sam Altman, executive director of the incubator startups Y Combinator. The sponsors include a number of influential figures in Silicon Valley, including businessmen Peter Theale and Jessica Livingston. The organization aims to create safe (that is, publicly available and open) Artificial Intelligence.



In December 2016, OpenAI introduced the Universe middleware for training and training strong AI. Theoretically, training can occur on all information of mankind, accessible through the Internet. These are games, websites and other applications.



OpenAI believes that reinforcement learning is an important way of machine learning that will greatly improve AI. In the process of learning by this method, the subject system (agent) learns by interacting with a certain environment. In contrast to traditional teaching with a teacher, reinforcement signals are a response to AI decisions, while some reinforcement rules are dynamically formed and difficult to understand for a person, that is, they are based on the simultaneous activity of formal neurons.



“Our ultimate goal is to develop a single intelligent agent who is able to flexibly apply the experience gained in the Universe to solve new problems and quickly gain new experience, which will be an important step towards a strong AI,” the OpenAI statement said.



Now the development of AI engaged in commercial corporations, such as Google, Facebook and Microsoft. Of course, they put their financial gain above the interests of humanity. The AI ​​created by them will act accordingly. OpenAI non-profit organization with an open-source alternative to AI is trying to resist corporations. All research within the OpenAI Institute is published in the public domain. In the official announcement of the foundation of the organization it is said: “In connection with the unpredictable history of AI, it is difficult to predict when the AI ​​of the human level can appear. When this happens, it will be important for mankind to have a leading research institute that is able to prioritize the gain for all over their own interests. ”