At Valve’s yearly DOTA 2 tournament - The International – an artificial intelligence beat one of the world’s best players in a pair of one-on-one matches. The bot, which is the work of Elon Musk-backed startup OpenAI, had already beaten some of the game’s top players before adding Danylo “Dendi” Ishuti to its list of conquests.

Engineers said the AI mastered DOTA 2 by playing games against itself, called “learned bot behavior,” allowing it to acquire techniques like last hitting and raze faking.

“We’ve coached it to learn just from playing against itself,” said OpenAI researcher Jakub Pachocki. “So we didn’t hard-code in any strategy, we didn’t have it learn from human experts, just from the very beginning, it just keeps playing against a copy of itself. It starts from complete randomness and then it makes very small improvements, and eventually it’s just pro level.”

OpenAI’s intelligence was able to beat regular DOTA 2 bots after an hour of learning. Reaching a level where it could beat pros like Dendi took just two weeks.

The exhibition match was supposed to be a best-of-five series, but Dendi threw in the towel after the AI scored a kill early during the second game. "I'm giving up," he said. "I don't think I'm getting it back. It's over."

While these 1-on-1 matches aren’t as complex as regular DOTA 2 games, OpenAI’s next step is for a full team of its bots to play in 5-vs-5 matches against the pros. Ultimately, it wants to mix the AI together with humans in a single team.

The OpenAI initiative is a non-profit dedicated to advancing artificial intelligence in a way that benefits humanity. It’s gained financial support from the likes of Amazon Web Services, Sal Altman, and Elon Musk, who is constantly warning us about the risks of AI.

The Tesla boss celebrated the DOTA 2 victory on Saturday, tweeting that eSports are much more complex than games like chess and Go. He followed this up with another tweet that warned AI was “vastly” more dangerous than North Korea.