Slashdot Mirror

← Back to Stories (view on slashdot.org)

Google Has a New Plan for China (and It's Not About Search) (bloomberg.com)

Posted by msmash on from the new-bet dept.

An anonymous reader shares a report: More than seven years after exiting China, Google is taking the boldest steps yet to come back. And it's not with a search engine. Instead, Google's ingress is centered around artificial intelligence. The internet giant is actively promoting TensorFlow, software that makes it easier to build AI systems, as a way to forge business ties in the world's largest online market, according to people familiar with the company's plans. It's a wide pitch targeting China's academics and tech titans. At the same time, Google parent Alphabet Inc. is adding more personnel to scour Chinese companies for potential AI investments, these people said. "China is a tremendous opportunity for any company because it is by far the single largest homogeneous market," said Kai Fu Lee, who headed Google's China operations before the company left in 2010. The market dwarfs any other, given how many Chinese people are online, and data from that "can be used to advance products, especially those relating to artificial intelligence," he added.

37 of 59 comments (clear)

  1. AlphaGo by Anonymous Coward · · Score: 1

    Unlike the US, the Chinese were blown away by AlphaGo's win against KeJie. Go is not widely played but universally known as a tough and tricky game.

    1. Re:AlphaGo by 110010001000 · · Score: 1, Insightful

      It isn't that impressive. Go is a game played with strict rules. Computers LOVE rules. They excel at anything that has a strict rule set. That is what they are best at. Why would it be a surprise that a computer can win at ANY game? They can win any game you invent.

    2. Re:AlphaGo by knightghost · · Score: 1

      Sounds like TensorFlow will love supporting the Chinese version of Communism and controlling all their serfs.

    3. Re:AlphaGo by AC-x · · Score: 2

      Go is a game played with strict rules. Computers LOVE rules.

      Go is a game with a huge search space. Computers HATE huge search spaces.

    4. Re:AlphaGo by Paradise+Pete · · Score: 1

      They can win any game you invent.

      They still suck at bridge, because players have to infer so much. Yes, Go is different because there's complete information, but what's really impressive about the Go dominance is that it came so suddenly. Before Alpha Go no program could challenge even strong club players, and the approach to playing like chess is a failure at Go. Amazingly, the difference in strength between the computer and top human is now much greater in Go than it is in chess.

    5. Re:AlphaGo by fluffernutter · · Score: 1

      Computers HATE huge search spaces.

      Until their processors grow large enough to deal with them. That is just an evolutionary problem.

      AI won't be revolutionary until it can deal with the exact opposite kind of problem; where rules CANNOT be explicitly defined.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    6. Re:AlphaGo by fluffernutter · · Score: 1

      The ruleset is 1/10 the size in Go as it is with Chess with the only problem being the size of the search space. So what you say isn't really a surprise at all. You pick a game with simple rules and figure out how to deal with the search space, you excel much faster.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    7. Re:AlphaGo by 110010001000 · · Score: 2

      No they don't. Computers LOVE huge search spaces. Human brains don't. You are confused.

    8. Re:AlphaGo by 110010001000 · · Score: 1

      If some smart programmers put effort into it they could eventually create a bridge playing program that would win. Playing games isn't AI.

    9. Re:AlphaGo by Anonymous Coward · · Score: 1

      AI can handle any situation where you have some input and you want specific output and you can somehow tell if the output is correct or not. That is the only rule the computers need. Typical example is where input is an image and output is text describing what object is in the image.

      What is interesting in AI is that lately they have used AI to decide if the output is correct or not and thus running an AI against AI. This has made it possible to generate AI that can draw realistic birds based on text description of what the bird looks like.

    10. Re:AlphaGo by 110010001000 · · Score: 1

      Because the sun is going to go supernova evenutally. You don't need to brute force search every problem. Just because some solution doesn't use brute force doesn't make it suddenly "AI".

    11. Re: AlphaGo by WindBourne · · Score: 1

      All games have rules. The more rules, the smaller the search space. Go is much harder for computers than chess due to it's simplicity and ability to set the board size bigger.

      --
      I prefer the "u" in honour as it seems to be missing these days.
    12. Re: AlphaGo by WindBourne · · Score: 1

      Wrong again. When you have small game with all search space, then no AI. When it is bigger than what computer can do in time, than all you can use is AI.

      --
      I prefer the "u" in honour as it seems to be missing these days.
    13. Re:AlphaGo by AC-x · · Score: 3, Insightful

      Until their processors grow large enough to deal with them. That is just an evolutionary problem

      The search space for Go is estimated at 10^170. If the world's fastest supercomputer could check one move every operation (93 PFLOPS) it would take 3.4*0^145 years to exhaustively solve it (at which point most physical objects will have decayed to subatomic particles).

      The fact that a relatively modest computer has beaten a human at Go today is impressive.

    14. Re:AlphaGo by AC-x · · Score: 1

      But the way they "solved" Go was AI...

    15. Re:AlphaGo by AC-x · · Score: 2

      The technically correct answer is computers WILL SIT THERE AND TRY huge search spaces, but human brains will HATE waiting forever for the answer.

    16. Re:AlphaGo by Anonymous Coward · · Score: 1

      Yea it kinda does
      If it's not brute force then its either intelligent in some way or random.
      You think computers win by randomly trying things and never learning from experience? Well the other option is intelligence. That intelligence wasn't programmed by a person, it's as artificial as it gets.

    17. Re:AlphaGo by ShanghaiBill · · Score: 1

      Because the sun is going to go supernova evenutally.

      Nitpick: The sun is too small to supernova. In roughly 5 billion years, it will expand into a red giant, the solar wind will blow away the outer layers until only a white dwarf is left, and then it will slowly cool.

      You are correct that we only have 5 billion years, since the earth will be destroyed during the red giant phase. Even if no one trips over the power cord, that is not enough time to brute force Go.

    18. Re:AlphaGo by ShanghaiBill · · Score: 1

      Playing games isn't AI.

      AlphaGo doesn't just play games, it also teaches itself how to play. The version of AlphaGo that beat Lee Sedol used a dataset of human games to train itself. But the latest version uses no human input other than the rules. It taught itself from scratch, learning not only strategy, but also rediscovering "joseki" tactics, and then going on to discover new tactics that humans never learned. It then beat Ke Jie, widely considered the world's strongest player.

      After his 0-3 defeat by AlphaGo, Ke Jie studied the program's behavior, and then went on a 20 game winning streak using the new insights, which is impressive even for someone of his standing: Statistically, he should have won only 60% of the games.

    19. Re:AlphaGo by fluffernutter · · Score: 1

      The computer doesn't have to see everything, it just has to see further than the human. The fact that it can is not a surprise at all.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    20. Re:AlphaGo by fluffernutter · · Score: 1

      Yet they say AI can do important things such as driving, where there is never a correct answer. Or at the very least there will be conflicting correct answers. If an AI drives through a construction site because its map said to follow the road, is it correct?

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    21. Re:AlphaGo by AC-x · · Score: 1

      With a branching factor of 235 per move (7x more than Chess) and a large typical lookahead by human players, even that's difficult.

      There is a reason why a computer beating a top human Go player was considered a big deal...

    22. Re:AlphaGo by WallyL · · Score: 1

      That sounds like a great article heading: "Your human brain will HATE waiting forever for the computer to provide this answer..."

  2. Google is back by hcs_$reboot · · Score: 1

    ...with a vengeance!

    --
    Slashdot, fix the reply notifications... You won't get away with it...
    1. Re:Google is back by Ritz_Just_Ritz · · Score: 1

      They're back with a vengeance until a Chinese government official (or group of them) invests in a competitor...at which point google will get run out of town again. Someone's gotta keep all that bitcoin flowing out of the country so it can get converted to proper currency and deposited in banks outside of China....

    2. Re: Google is back by WindBourne · · Score: 1

      Worse yet, all of Google's tech ip will be transferred to China. The current CEO of Google is a true idiot.

      --
      I prefer the "u" in honour as it seems to be missing these days.
  3. AI by 110010001000 · · Score: 1

    More AI hype. Next thing you know they will be naming hard drives "AI" because it is so cool and trendy. AI is the next VR.

    1. Re:AI by Baron_Yam · · Score: 1

      >AI is the next VR.

      VR should definitely go away until we can produce a direct neural interface (I don't see solid holograms and force fields ever being a thing, so no holodecks for us).

      On the other hand, I'm quite enthusiastic about AR. I already wear glasses anyway, and would not mind a computer generated overlay for reality.

    2. Re:AI by 110010001000 · · Score: 1

      AR definitely has potential. VR has issues that cannot be overcome due to human physiology.

  4. For a change by Anonymous Coward · · Score: 1

    This time, we don't censor results. We just sell an AI that will censor instead.
    capcha: predicts

  5. Let's see what they've learned by rmdingler · · Score: 4, Insightful
    In its best light, the domestic Chinese market is protectionist and prone to imitation beyond the scope of copyright enforcement.

    Alphabet must now believe they can make inroads to a worthwhile share of the admittedly gigantic market without ultimately losing their technology to domestic companies.

    So the $64,000 Question is, "What sort of arrangement can they (have they) made with Chinese officials that leads them to believe Lucy won't pull the football away at the last second?"

    --
    Happiness in intelligent people is the rarest thing I know.

    Ernest Hemingway

    1. Re:Let's see what they've learned by Anonymous Coward · · Score: 1

      Tensorflow is open source. Can you send my $64,000 to my email please. See info from my profile.

  6. Google plan to sell people out by WCMI92 · · Score: 1

    Because of the money communist slave labor doesn't have that they want.

    I am all for making money. But not at the expense of victims of communism.

    --
    Corporatism != Free Market
  7. Company wants to sell stuff into a large market... by frank_adrian314159 · · Score: 2

    News at eleven.

    --
    That is all.
  8. Re:Rome to Barbarians by AC-x · · Score: 2

    China has been making most of our high technology products for quite a while now...

  9. Re:"Strip mine their privacy for money" is not new by crimson+tsunami · · Score: 1

    It's so cute that you think you do.

  10. Not about searching for by AHuxley · · Score: 1

    Tiananmen Square, 1989 https://en.wikipedia.org/wiki/...
    89 student movement, 89 student strike, 89 people's movement.. all the people topple communism, impose martial law, Tiananmen Mothers movement, Victoria Park, Operation Yellowbird https://en.wikipedia.org/wiki/.....
    A search engine that fails to search :)

    --
    Domestic spying is now "Benign Information Gathering"