OpenAI Bot Conquers Humans at ‘Dota 2’

Another day, another human champion brought low by an artificial intelligence (A.I.) platform.

This time, the contest in question was “The International,” a huge tournament centered on “Dota 2,” a multiplayer video game in which virtual heroes battle for control of a battlefield. Fighting on behalf of A.I. platforms everywhere: a bot created by OpenAI, a non-profit A.I. research company. Its opponents: a collection of top-ranked “Dota 2” players, including Arteezy (a.k.a. Artour Babaev), who is widely viewed as the best on the planet.

Thanks to machine-learning algorithms that allowed it to quickly adapt to the game’s fast-moving environment, the bot destroyed all human opponents. According to GeekWire, one challenger, Danylo “Dendi” Ishutin, refused to play a third match against the bot after two rough rounds. Here’s a YouTube clip of his whipping:

“Dota 2” creator Valve touts “competitive balance” as the game’s “crown jewel.” That balance also makes it perfect for competitions between human beings and autonomous software; if the game gives neither side an undue advantage, then it should (theoretically) serve as the perfect proving ground for individual mastery—or individual destruction at the hands of an increasingly sophisticated bot.

“Agents must learn to plan, attack, trick, and deceive their opponents,” OpenAI wrote in a blog posting recounting the competition. “The correlation between player skill and actions-per-minute is not strong, and in fact, our AI’s actions-per-minute are comparable to that of an average human player.”

Elon Musk, one of the co-founders of OpenAI, issued several tweets about the competition, including one in which he thanked Microsoft “for use of their Azure cloud computing platform.” He also suggested that competitive e-sports, in which groups of players battle for video-game supremacy, is “vastly more complex than traditional board games like chess & Go.” (A.I. platforms have already mastered those latter two games.)

Shortly after praising the OpenAI bot, Musk returned to Twitter to frame the dangers of A.I.: “Nobody likes being regulated, but everything (cars, planes, food, drugs, etc0 that’s a danger to the public is regulated. AI should be too.”

Indeed, OpenAI’s stated mission is to create A.I. that is “safe.” In addition to building bots, it also releases papers and open-source software tools for A.I. research. “We will not keep information private for private benefit,” the organization stated at one point, “but in the long term, we expect to create formal processes for keeping technologies private when there are safety concerns.”

While OpenAI hasn’t open-sourced the bot that allowed it to win so handily at “Dota 2,” the organization does offer a variety of tools, including open-source software for robot simulation, as well as a framework for developing reinforcement learning algorithms. If you’re interested in A.I. and machine learning, its portfolio is well worth checking out.

Post a Comment

Your email address will not be published.