“#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,” Gates tweeted Tuesday.
Individual robots have previously been able to win a solo version of Dota, says Christy Dennison, an Open AI machine learning engineer, in a video published Monday. But team play is different.
“Dota is one of the most popular and challenging competitive video games ever. Playing Dota means you have to coordinate and focus as a team of five. Last year, we built a bot that defeated the world’s best players at the one-[versus]-one minigame. This year, we want to beat the best pro teams at the full game. To do this, we have built a new AI system which we spent the last few months training,” Dennison says.
“We are now starting to play against amateur teams to test our skill and to our surprise, so far, we have won our first games against every team we have tested,” Dennison says.
The robot teams learn to play Dota by playing against themselves, according to a statement by OpenAI. The robot team, OpenAI Five, is able to play 180 years worth of games each day, according to the statement.
The rules of Dota are to kill the opposite team and take the buildings, according to gamer William Lee, known in the Dota community as "Blitz." The bots beat the humans with some restrictions to play, OpenAI says — some elements of play, such as invisibility, summons and illusions, were not included.
Still, Lee, who watched the bots play, is impressed with the robots' play strategy.
"This is one of the highest level plays that you can make,” Lee says in the OpenAI video. “The ability to intuitively do this is insane. Doing it one game, I could maybe chalk it up to dumb luck, doing it two games in a row ... that it is more than just coincidence,” says Lee.
“I would say — it was pretty easy to quantify for me – it was about eight years for me to learn the strategies that the bot was intuitively doing,” he says.
The development of the bots ability to compete against human teams playing Dota is exciting for the applications of artificial intelligence beyond playing video games, says Dennison.
“Overall what we are excited about is that the training method we use is very general. We are focused on learning Dota, but we are hoping that this will give us more and more insight about how AI can solve complex problems of any kind,” Dennison says.
Gates is supportive of the aim of OpenAI to develop artificial intelligence for good.
“This is just one of many amazing projects I had a chance to see at OpenAI, where they’re working to ensure as many people benefit from AI as possible. This is an incredibly important mission, and I’m excited to see more of their work,” says Gates on Twitter.
This is just one of many amazing projects I had a chance to see at @OpenAI, where they’re working to ensure as many people benefit from AI as possible. This is an incredibly important mission, and I’m excited to see more of their work.
The OpenAI team will continue to develop the Dota robot team with an eye towards beating professionals at the world championship of Dota, The International, in August.
“We are still far away from beating pro teams but I think everyone here was surprised to see this,” says Dennison of the success the bots have already achieved.
Ultimately, the goal of OpenAI, which launched in 2015, is “discovering and enacting” a safe future with artificial intelligence, according to the nonprofit’s website. Elon Musk is a co-founder of OpenAI but left the nonprofit’s board in February to “eliminate a potential future conflict” as his company Tesla becomes more involved in artificial intelligence, Open AI said in a written statement. After his departure from the board, Musk would “continue to donate and advise the organization,” Open AI said at the time.
Like this story? Subscribe to CNBC Make It on YouTube!