The Dota 2 esports tournament in Seattle, known as The International, demonstrates how video games have become hugely popular team sports — and today alpha geek Elon Musk took the occasion to tout a different kind of team effort.
Musk, the CEO of SpaceX and Tesla, is also a big supporter of OpenAI, a nonprofit research company aimed at boosting artificial intelligence architectures and applications that benefit humanity.
Today, OpenAI showed off an AI bot that vanquished Danylo “Dendi” Ishutin, one of the world’s top Dota 2 players, in a one-vs.-one demonstration match.
“OK, this guy is scary,” Ishutin said as he battled the bot’s minions. The crowd at KeyArena groaned when the bot crushed Ishutin’s game avatars.
Ishutin was beaten badly in the first match, forfeited a second match, and refused to play a third.
He’s not the only player to feel the burn.
“Over the past week, our bot was undefeated against many top professionals including SumaiL (top 1v1 player in the world) and Arteezy (top overall player in the world),” OpenAI said in a blog post recapping the action.
The Open AI bot built up its video-game skills not by processing the results of human-vs.-human contests, but by playing itself over and over again.
“This is a step toward building more general systems that can learn really complicated, messy and important real-world tasks, like being a surgeon,” OpenAI’s co-founder and chief technology officer, Greg Brockman, said after the match.
In Musk’s eyes, OpenAI’s win was an achievement bigger than Google DeepMind’s development of AlphaGo, the AI bot that has bested the world’s top players of the ancient game of Go:
OpenAI first ever to defeat world's best players in competitive eSports. Vastly more complex than traditional board games like chess & Go.
— Elon Musk (@elonmusk) August 12, 2017
Musk paid tribute to the OpenAI team, and also gave a shout-out to Microsoft Azure, which provided the platform for the Dota 2 bot:
Would like to express our appreciation to Microsoft for use of their Azure cloud computing platform. This required massive processing power.
— Elon Musk (@elonmusk) August 12, 2017
But tonight’s Twitter conversation wasn’t all fun and games: Musk engaged in a back-and-forth discussion about the potential risks posed by AI – a topic he’s often touched on. Along the way, he picked up on a reference to Iain M. Banks, the science-fiction author who wrote sentient starships into his novels:
If you're not concerned about AI safety, you should be. Vastly more risk than North Korea. pic.twitter.com/2z0tiid0lc
— Elon Musk (@elonmusk) August 12, 2017
On the other hand, maybe Iain M Banks was right…
— Iain Thomson (@iainthomson) August 12, 2017
I sure hope so, but we can’t take that for granted
— Elon Musk (@elonmusk) August 12, 2017
Nobody likes being regulated, but everything (cars, planes, food, drugs, etc) that's a danger to the public is regulated. AI should be too.
— Elon Musk (@elonmusk) August 12, 2017
AI on its own will not be a danger to humans for the foreseeable future. AI combined with weapons, maybe. So, regulate weapons, not AI
— Reza Zadeh (@Reza_Zadeh) August 12, 2017
Biggest impediment to recognizing AI danger are those so convinced of their own intelligence they can’t imagine anyone doing what they can’t
— Elon Musk (@elonmusk) August 12, 2017
agree with regulating such dangerous things, how do you regulate something that can be written by a lone hacker late at night in a garage?
— Vandy (@VeryVandy) August 12, 2017
It is far too complex for that. Requires a team of the world’s best AI researchers with massive computing resources.
— Elon Musk (@elonmusk) August 12, 2017
To top it all off, Musk joined in on a “Game of Thrones” riff:
Elon Musk : AI :: John Snow : White Walkers
— Jason Toff (@jasontoff) August 12, 2017
Wintermute Is Coming
— Kevin Boyd (@Beryllium9) August 12, 2017
For sure
— Elon Musk (@elonmusk) August 12, 2017
Speaking of White Walkers … creating AI bots capable of beating humans at Warcraft-style video games might seem just a little too close for comfort. Maybe OpenAI should be working on virtual dragonglass instead.