What just happened? While we’ve already seen artificial intelligence bots best human opponents in Dota 2, they weren’t good enough to beat one professional team at the game’s annual The International tournament. But with two more matches to go, that could change.
Earlier this month, bots created by OpenAI, a research non-profit founded by Elon Musk, competed against four pro Dota 2 players and one commentator in a best-of-three exhibition series. They had been taught to play the MOBO using machine learning techniques that were the equivalent of 180 years of playing per day. That training looked to have paid off when they won convincingly.
As was the case with Google’s AlphaGo, it was starting to look as if another artificial intelligence had been created that could beat humanity’s best. But the bots have since lost to other players, and Brazil’s Team paiN, who are currently ranked 14th in the world, also showed that AI still isn’t yet able to dominate in Dota 2.
The match, which lasted 45 minutes, was pretty close, and the bots gained more kills than the pros following a stuttering start. But, as noted by The Register, the final battle saw the AI players wiped out, leaving their Ancient exposed, at which point Team paiN destroyed it for the win.
“We were definitely less confident about winning this time round, because the pros are a lot harder to face. It’s disappointing to lose, but I’m happy that we could stand against a top team in such a long match,” Filip Wolski, a member of technical staff at OpenAI, told the publication.
The AI reportedly made some strange decisions during the match, such as casting powerful spells when few or no opponents were nearby.
While the OpenAI Five team lost this match, it still has two more games against other pro teams in which to redeem itself.
In other Dota 2 news, check out Valve founder and CEO Gabe Newell recording his voice for the game in this hilarious sketch.