Slashdot Mirror
China Censored Google's AlphaGo Match Against World's Best Go Player (theguardian.com)
DeepMind's board game-playing AI, AlphaGo, may well have won its first game against the Go world number one, Ke Jie, from China -- but most Chinese viewers could not watch the match live. From a report: The Chinese government had issued a censorship notice to broadcasters and online publishers, warning them against livestreaming Tuesday's game, according to China Digital Times, a site that regularly posts such notices in the name of transparency. "Regarding the go match between Ke Jie and AlphaGo, no website, without exception, may carry a livestream," the notice read. "If one has been announced in advance, please immediately withdraw it." The ban did not just cover video footage: outlets were banned from covering the match live in any way, including text commentary, social media, or push notifications. It appears the government was concerned that 19-year-old Ke, who lost the first of three scheduled games by a razor-thin half-point margin, might have suffered a more damaging defeat that would hurt the national pride of a state which holds Go close to its heart.
Sorry. you guys just made that way too easy!
BeauHD. Worst editor since kdawson.
I watched it live on WeChat... do I need to feel worried now?
...to your local Party HQ for sentencing. You are required to provide your own bullet.
MCTS programs don't care about the winning margin. It was quite clear that Ke Jie was behind, but AlphaGo just didn't take unnecessary risk to win by a large margin.
"...that would hurt the national pride of a state which holds Go close to its heart."
Perhaps we should remind the country that we're talking about a game here.
Hell, Kasparov lost to Deep Blue 20 years ago. The concept of a world champion being defeated by a computer playing a game ain't exactly new.
Because if you didn't see it, it didn't happen.
Weaselmancer
rediculous.
I'm trying to find where to watch an English stream, but all I'm finding is news about China's censorship. That's great, but it's affecting me by proxy because I can't find somewhere that's going to stream it here!
The second match is at 0330 UTC on Thursday (late evening today, Wednesday, in the US)... where will it be broadcast?
"I will trust Google to 'do no evil' until the founders no longer run it." Hello Alphabet.
Do no watch GO, do not collect $200.
What would be truly scary is if the computer dragged the game out while taking over various infrastructure systems, and then simultaneously with the winning move electrocuted the wife and kids of the player, emptied his retirement accounts, and erased his name from the citizenship roles.
I find it interesting that just a few days ago there was an uproar with respect to the Univ. of Maryland commencement speech by a Chinese student (studying abroad here in the US) on freedom of speech. There was a huge backlash from China. And now this...
Where else would we get all of our cheap plastic crap?
What if the AlphaGo AI is so advanced, that it is toying with its opponent, and letting him near-win? What if it has the strength of "the Hulk" with the subtlety of Black Widow?
This is completely misunderstanding the nature of "soft AI" contained in Alphago. You're clearly talking about hard AI, which AlphaGo is not.
AlphaGo doesn't doesn't know that there is a world outside Go. It doesn't learn. It doesn't know the meaning of 'taunt.' It doesn't know it's playing a game against the world's top human. It can't self-introspect, it can't change its programming, and it doesn't know a Go board looks like, and doesn't even know what Go is.
AlphaGo is a classifier. Given a set, it will score potential additions to the set based on previous data. It also has some code to coordinate that with a Go board, but that is mundane code, it's not even soft AI.
"First they came for the slanderers and i said nothing."
AlphaGo keeps track of both the score and of the statistical likelihood that it'll win the game. Being able to constantly evaluate these two values with high accuracy is actually a big competitive advantage. As the score doesn't really matter, it is entirely by design that during the end-game, AlphaGo deliberately sacrifices score points in order to gain a stronger position and increase the likelihood of an overall win.
In some way, winning with a minimal score demonstrates better control of the game than winning with an arbitrary and larger score. That just means you needlessly took risks that you didn't have to.
Explain exactly why the Chinese government should not be thrown out, and/or why the world should stop doing business with China until that happens?
Because (a) doing business with China is profitable, and profits trump morals; and (b) China, with its stronghold on manufacturing, has "the world" by the balls.
The reality of the modern age of computing...computers are going to be better at a lot of games than humans. Period. Just accept it now. Some games may take longer to match and exceed human capability, but it will eventually happen.
Welcome to reality,China.
Bearded Dragon
I guess Beijing didn't want any weiqileaks.
Why can't we go back to using jumpers to configure slot adapter cards? Why? I say!
It goes both ways. (c) China is the worlds most populous country, and due to rapid industrialisation it is full of people suddenly finding themselves several times wealthier than their parents could have dreamed. Non-Chinese companies want a slice of that pie.
China doesn't value free speech in the same way that is typical in America. They do very strongly value unity, and stability, and social cohesion.
> What if the AlphaGo AI is so advanced, that it is toying with its opponent, and letting him near-win? What if it has the strength of "the Hulk" with the subtlety of Black Widow?
The algorithm is neither. As discussed frequently, it pursues the highest probability path to a win, not the highest winning margin. This means taking less risks when apparently ahead, which may look like "letting him near win" and "toying with him", by giving up small amounts of territory as long as those preclude lines of action by the opponent which would potentially open up more. In US football, consider giving up run yards in 4th quarter to preclude passing touchdowns.
There is no programming regarding an opponent's predicted emotional states.
AlphaGo doesn't try and maximize its win margin. It would have won by 15 points or more if the winning margin mattered.
When all paths lead to victory there is numerical instability in the rollouts so a move that gives a 15 point win margin, might, by chance get say 99.995% chance of winning, but one of the billions of other paths that also lead to a win will, by chance - give a rollout of 99.996% chance of winning. So every move in a won game is essentially random and will tend to reduce the win margin against a skilled opponent (who will always make a move that decreases their loss margin) until the win margin is 1/2.
a more damaging defeat that would hurt the national pride of a state which holds Go close to its heart.
Does anyone else see a societal pressure point ripe for needling with social media bots?
The Russians have won. They have made the world a cesspool of distrust, greed, fear and hate.
While it is to an extent censorship, it is also their temporary measure to stop a potentially deadly snowball effect.
Surely there aren't any Professional Go players on slashdot, but do imagine what happen if you've putted in 30 years for a job and found out today it is irrelevant? You will be in despair and so will your colleagues.
It's not healthy to cause a chain reaction that result in no more new Go players being encouraged to join the competition.
In fact, I highly doubt even google's researchers have any idea what they have done. They've made an AI great at winning, they've already proved AI potential. Now what? Winning all the world's game and destroying the Go's market? Go players play to win, everyone else supports them to see a competitive game. Google's AI encourages neither. It wins all of them (no player gets to win), and wins too much (3-0 again, boring).
If Google wants AI to be in Go's competition, it needs AI vs AI competition. At the current state, even China knew AI vs Player could potentially destroy their markets, so for them to do something to pause against the effect is somewhat expected.
Did not work so well in Iraq, did it?
If the Chinese government would simply be thrown out, the country would collapse in to 3 or perhaps even 10 nations. The north Koreans would lose the only lash they are kept in control with. Some war lords, generals who have access to the right weapons, would found mini dictatorships. The nuclear weapon arsenal would be out of control. Millions of Chinese would emigrate to the smaller asian nations like Vietnam, Malaysia and Thailand.
Should I continue?
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.