Yesterday I reported on how a mystery account called "Master" had been tearing up the world of online Go, and nobody knew who it was. But in an update posted on Twitter earlier this morning, Google has admitted that the mystery account "Master", as well as a second account "Magister", was their world-beating AlphaGo bot all along.
The revelation came courtesy of a statement posted by Demis Hassabis, co-founder and chief executive of the Deepmind artificial intelligence company. Deepmind was bought out by Google in 2014, and the same company is responsible for building the machine learning algorithms that powered the AlphaGo AI - the latter of which became the first Go-playing bot to defeat a professional Go player last year.
Excited to share an update on #AlphaGo! pic.twitter.com/IT5HGBmYDr
— Demis Hassabis (@demishassabis) January 4, 2017
The revelation is only partial vindication to the Go community: while the general consensus was that "Master" was an AI - who racked up 51 wins straight before being forced into a tie on its 52nd game thanks to an old-fashioned disconnect - people didn't believe Google was responsible.
After all, they were the same company who made such pomp and circumstance out of the five-match series against Lee Sedol. And as of yesterday, AlphaGo was the second highest ranked Go player in the world. Surely, if Google was going to unleash something that powerful on online Go, they wouldn't do so without saying anything.
But as it turns out, that's exactly what Google did. It really is a bit like that episode from Hikaru No Go, where Hikaru spends time in an internet cafe so his ghost mentor and legendary Go champion Sai can play matches to his heart's delight.
Perhaps the most lovely part of all of this has been the reception from the Go community. Rather than acting with shock and dismay towards Google, the general response has been one of excitement - possibly partly because people can witness more matches featuring AlphaGo, giving everyone a new world of moves and strategies to study and learn.
It's the real deal. AlphaGo is probably absolutely unbeatable in its current state. #baduk #go #weiqi #igo https://t.co/QyhPiEYcA5
— Sherico Dela Cruz (@Mithfalath_) January 4, 2017
Ke Jie, the current #1 ranked player in Go who fell victim - twice - to AlphaGo's online rampage, posted on Weibu that he was aware of AlphaGo's identity. "Thanks for the shock and awe the latest version of AlphaGo provided us," he wrote, according to a fan translation.
"As someone who knew its identity from the beginning, I'd have loved to see us human winning a blitz game. If not for being in the hospital, I'd have tried one last response that I prepared for a week...slight regret, and hope Gu Li can play to the best of our human ability in this final game of AlphaGo beta test."
AlphaGo is due to play more official matches this year, although the next step is to see how well AlphaGo plays under long time constraints. (The matches played on the online Korean and Chinese servers were under blitz time formats, with AlphaGo making all of its moves in less than 10 seconds.) Nonetheless, you'd have to favour AlphaGo in any format right now - and it's fascinating to watch, regardless of the format.
Comments
Not really a surprise. I imagine that Google was keeping it quiet because they either didn't want a flood of people challenging the machine or they wanted to see if people could work out that it was an AI. Kind of like a Turing test for board game AI.
Join the discussion!
Comment Voting
Up Votes
Down Votes
Only logged in users may vote for comments!
Please log in or register to gain access to this feature.
Get Permalink