Perhaps this is because I don't have the hard facts and figures right in front of me, but I have a hard time believing that foreign tech investments in China really pay off. The Chinese strategy seems to be to woo big foreign companies, build a critical mass of local users who want the product, ban or otherwise limit the foreign company's influence in China, then develop a home-grown Chinese version of the foreign company's product with state support and the newly acquired domestic talent.
If Google accomplishes anything worthwhile at this research center, won't China just shut it down and re-employ those researchers at a domestic company? Why do high-tech companies even continue to try to break into China?
An AI center is not a product like a search engine or Windows, or an e-commerce site.
It helps China cultivate talents, why would China limit its influence.
MSRA accomplishes many things worthwhile such as ResNet in China, and it prospers now. Some researchers enter MSRA, and some leave. They enter or leave not because the government force them to do so, they enter or leave because they can tell where opportunity resides.
I'm not asking why China would limit it, I'm asking what Google gains from it.
If Google's top talent in China will mostly be poached within the next 10 years, is it really worthwhile to open a research center there? MSRA is a very good example, but at the end of the day, it's still a training ground for Chinese researchers. Paradoxically, the more important it is that Chinese researchers work on at Google China, the more likely that Google is to lose those researchers to government or private competitors.
I made a video that actually analyzes the data behind this. Basically the entry-level of the field is about 50% Chinese and the more experienced levels are about 10-20% Chinese. On the other hand only perhaps 5% of work actually comes from Chinese labs.
Not necessarily easier, just cheaper. Google can throw as much money at ML as it wants, but I contend that a top ML researcher is probably worth at least 3-5x more than mediocre one.
Yeah, I've worked with quite a few top ML researchers at US universities and many of them are Chinese nationals. Even here in the US. With China's economy booming the way it is and the overall political climate in the US right now, many see a better quality of life for themselves and their families in China. It's not just about cost.
What are they making? I've not heard of any >$200k/year ML researchers returning to China: usually it's good, but not amazing, researchers that do return.
check the board of directors and the top management team in Google, can you see any Chinese name in the list? now check the same lists for BATJ, Alipay and Huawei.
with such bamboo ceiling in place, google doesn't stand a chance to complete for the best of the best.
On one hand, I have to agree that China's closed/censored society is directly at odds with the product offering of a company like Google ... on the other hand, does one simply disengage from such a massive market as China? Perhaps Google is betting on a future when there is greater freedom for the Chinese people.
I personally think Google's best approach would be to pay hand-over-fist for the best Chinese AI/ML experts to emigrate to the US and do their work here. However, Google seems to be taking the shotgun approach of employing more experts in the short-term while guaranteeing them less job stability in the long-term.
I think you're right in that it really does depend on when China becomes a free society. Personally, from what/who I know from China, I don't see this happening within the next decade. I expect political instability within China to reach a tipping point, after which we'll either see a very slow global liberalization or a definitive move towards a more totalitarian Chinese government with a modern twist.
Honestly I don't know why Google has any faith in the good will of the future Chinese government. As has been established for a very long time, China needs foreign expertise more than any foreign company needs a labor or consumer market within China.
I think Google just wants to recruit AI talents from China and uses research from the AI center in China in markets it is not banned.
If so, they don't need to interact with the government that much like when they offered a search service to Chinese netizens.
I can think some companies that just do this. Hulu has an office in Beijing, although it doesn't offer video streaming service in China. And Grab has an office in Beijing too, although it only targets Southeast Asia market.
The basic strategy is to woo Western companies with extremely cheap labor and vague promises about access to the Chinese market, in order to get the companies to export most/all of their capital equipment and business expertise overseas. Then this is used as leverage to force them to adopt PRC-friendly policies (see Apple and government censorship in China, for example) and IP transfer to Chinese firms. Keep in mind that in China there is no objective judicial system as in the West to mediate disputes between the government and firms; once your assets are there you are at the mercy of the PRC.
It's really important to understand that this is a long-term geopolitical strategy by the PRC to supplant the US as the dominant global superpower, and a big part of it is siphoning off the industrial and manufacturing capacity of the west (see One Belt One Road Initiative).
Fair, but in my opinion a researcher who first publishes X result is usually the best equipped to first publish (or not publish, simply discover/use) Y result which depends on X. If we freely gave all US military research to Zimbabwe they wouldn't develop a top-tier military program because they don't have the expertise to understand and implement the knowledge that we have. This isn't the case with China
Isn't something published available to all the world?
If Google first publishes some result X, and Chinese researchers find this result, they are free to use this result to publish some result Y, regardless of whether Google opens an AI center in China.
This will sound condescending, but it's not meant to be,
The short answer is no. The field of research is completely fucked up. You are incentivized to explain your research in as much detail as is required to get a top publication (which honestly isn't much, just look at NIPS this year, there are a ton of overly vague, "exploratory" pubs that really describe jack-shit) but absolutely no further, as this could allow a competitor to fully continue your research and get the next major pub from it. So you want to describe enough so that people understand your research at a high level, but so that nobody can truly understand at a technical level what you've done.
So in essence no, research isn't something that is as freely communicable as publications. It really does require personal expertise, in the sense that there is a ton of knowledge contained in experts that you can't find anywhere in publications.
> but I have a hard time believing that foreign tech investments in China really pay off.
If it didn't pay off, apple wouldn't be the most valuable company in the world.
> The Chinese strategy seems to be to woo big foreign companies, build a critical mass of local users who want the product, ban or otherwise limit the foreign company's influence in China
What else do you expect them to do? Cede their entire market to foreigners? Every country did the same thing. The europeans did it. We did it in the US. The japanese and the south koreans. It's really common sense.
> If Google accomplishes anything worthwhile at this research center, won't China just shut it down and re-employ those researchers at a domestic company?
If they did, why would google invest in china? It's been 40 years. If investing in china wasn't lucrative, it would have ended a long time ago.
> Why do high-tech companies even continue to try to break into China?
The real question is why is it that we get the exact same type of comments in every thread about china?
Go through every china related thread on HN and every single one has the exact same comments. Why?
A lot of people still are questioning the motion. But it's fair and simple: Google needs to tap local Chinese AI talents. Their size is grow so large that it can't be ignore. And you don't have to read it between the lines, it's front and center on the Google's official blog post[1], stated by Google Chief Scientist for AI/ML Feifei Li:
China is home to many of the world's top experts in artificial intelligence (AI)
and machine learning. All three winning teams of the ImageNet Challenge
in the past three years have been largely composed of Chinese researchers.
Chinese authors contributed 43 percent of all content in the top 100 AI journals in 2015
and when the Association for the Advancement of AI discovered that
their annual meeting overlapped with Chinese New Year this year, they rescheduled.
Didn't expect a simple job offering would bring all these political arguments and speculation from people hardly read any Chinese or understand Chinese history & economics. smh. It's just business, China has a strong STEM talent pool, and Google can easily select them from top in top at a competitive price.
Please understand that we are still in the midst of a 'war for talent' as far as AI is concerned. Since getting them to the US is less likely, Google is just moving to where the talent is.
True. It is a pretty strong endorsement from Google to setup a specific research division in a country where they effectively get banned. The only logical reason is to hire talents at local price for projects outside of China. Not sure how the return would be though.
When google initially entered the Chinese market more than a decade ago, it offered a censored version of google search conformed to the censorship requirements set by the Chinese government. It argued that the most ethical option was to offer some restricted services to Chinese rather than leave hundreds of millions of Chinese Internet users with limited access to information [1]. Basically it argued that Chinese need google (and other American internet companies) to avoid being left behind in the Internet age.
Interestingly, 7 years after its quit from the Chinese market, the Internet sector in China is booming with 4 out of world's top 10 Internet companies being Chinese companies [2]. Now they are setting up this AI centre, surely that is to help Chinese avoid being left behind in the AI age again.
Google.cn was better than the alternatives, since it at least disclosed when search results had to be censored (sorta like the notices when links are removed due to copyright takedowns). Of course the CPC didn't like that, and it was probably one of the things which ultimately led to Google leaving/being driven out of China.
It would have been much more profitable to simply comply rather than lose the Chinese market, as numerous other Western companies have done.
I'm pretty sure that was, from the very beginning, just corporate lingo for "the profits outweigh the moral concerns". In retrospect, it was inevitable that China would develop its own Google/Facebook/Amazon through their strange neo-mercantillist economic strategy.
An under-reported fact is that Google has done more than any other country to help spread the use of AI in China. AlphaGo was a Sputnik moment that galvanized the government, which threw its weight behind AI a few months later. TensorFlow is widely used and Google is paying evangelists to tour the country, giving speeches and writing books about its use. This hiring center is just one piece of a larger strategy for them to ensure that TensorFlow wins the API wars in machine learning.
Please HN don’t censor this comment. I legitimately believe that the US and China are on the road to war, and this war will draw on in all the powers of the world.
This comment might be mocked, but if you read into history you can see how unbelievable war seemed within weeks of a conflict. The day WWI broke out the French and German ambassador were trying out each other’s cars and drinking wine in Alsace.
This research center is a bad idea. It’s like if the University of Chicago and Enrico Fermi decided to establish their atomic research lab in Munich in the 1930’s.
Don’t see this as an anti-Chinese comment, if anything the fault lies with Americans who abused, overused, and failed to adjust an international trade/monetary system. The fact that our government is now run by no one who served in or remembered World War II / The Great Depression seems a likely cause of how we went down this path.
If you’re curious for more on this, look into how close the US came to war with China in 1995 in Taiwan. Also look up the RAND Corporation’s study on war with China from 2015.
How many wars were started over tensions concerning research centers / science? WW1 was started by an assassination, WW2 by an aggressive annexation of countries and race-based extermination.
If China and the US go to war, it'll probably be over aggression in North Korea; not over a corporately funded scientific research center.
The parent comment is not claiming that war will start because of a research lab. (S)he is saying it is against the interest of the US to start an AI lab there because AI will be critical to the war effort if there is war.
Maybe that Google sets up an AI lab is not in the interest of the US, but it is in the interest of Google.
Google is a multinational company and its employees come from diverse countries. Although it is headquartered in the US and its founders are American citizens, its core interests may not fully align with the core interests of the US.
Why would the US go to war with China? What would we gain from it? The US would like real access to the Chinese market, but war isn't a good way to do that, because the Chinese population would refuse to buy anything from US companies. Additionally, the first thing that would happen in a war is that China would stop manufacturing things for US companies, which would massively disrupt the companies, probably cause a stock market crash (for whatever that's worth) because of lowered future profit, and create a shortage of consumer goods. We actually benefit from the way things are now, we get cheap consumer goods, and we get sort-of access to the Chinese market.
I kind of think war could happen, not for the reasons you listed, but because the odds of the US winning such a war might be good but will only go down with time, which is the kind of dynamic that leads to aggression.
Was it known the nazis were evil when IBM helped them? Did IBM even know what their stuff was used for?
Different times, different situations. The US was full of anti-semites too, and didn't hesitate to put anyone with Japanese heritage into concentration camps after Pearl Harbor - and nobody cared. That stuff has been covered up / hardly mentioned, ever since. And the US and its citizens doesn't seem to have a problem putting "enemy combatants" into concentration / torture camps after 9/11 (Guantanamo Bay). And that's just the stuff we know of. There's also the much more mysterious CIA dark sites, who knows what went on in there.
I'm not saying China is any better from a moral standpoint, but the US and the businesses in there - often working together - aren't any better. See also the Snowden leaks, that proved the big tech companies were in cahoots with the NSA, amongst other things to profile for "enemy combatants" in the US.
Edwin Black made a pretty convincing case that the IBM chairman Thomas J. Watson, Sr was quite aware of what the IBM equipment was used for - and actually quite enthusiastic about it all.
So think of that connection next time you use "Watson".
First, Chinese government is not Nazi and Google is not IBM. You can't just say that a government is Nazi just because you hate that government.
Second, even if Google openly broke with Chinese government, it still has offices in China. I doubt that Google opens an AI center in China to "allow its DeepMind technology to be used to spy on Chinese citizens and dissidents", and I do think that Google has a big ambition than you can think in China. And I think that ambition is to advance AI with the help of local talents.
There is a difference between an analogy and a comparison. Just think of it as it working with murderers for commercial gain, and try again, as far as the analogy is concerned.
China is plenty totalitarian, too, so even as a comparison it's not that far off (though you won't be able to follow comparisons either if you think things have to be exactly the same thing to be compared). You might even say the Nazis weren't Nazis before they started and lost their war; if they had stuck with, say, attacking Poland and Russia, and had kept their concentration camps secret (i.e. plausibly deniable), they still might be darlings of Western business elites even today, as Hitler was before the stuff we now have 20/20 hindsight on.
Last but not least, a big ambition is living a life you then don't have to lie to yourself about. Compared to that, "advancing AI" isn't even an ambition, nevermind a big one.
At its core, the goal of the two regimes (China, Nazi Germany) is the same: control information so that the existing regime is portrayed in a good light that preserves stability. Just because China isn't genocoding Jews doesn't mean they don't have hundreds of thousands of political prisoners, disadvantaged minorities, and a completely authoritarian government. As is, you would never hear of the North Korean concentration camps that China supports, nor the Tibetan and Uighur genocides that are ongoing.
Google's move is purely financial. Google's biggest ambition is to control Chinese search, but its greatest ambition is to control all search, and the internet as whole. Google, as all other appreciably large tech companies, wants to hold a dominant position in the Chinse market, and to do so is willing to sacrifice market share in inferuor products. How much China gets involved is TB.
The really fun part? Watching corporate entities justify genocide, sexism, racism, etc. in their grabs for dominance over the remaining European provinces.
>"The Tibetan government in exile in India estimates that the death toll since 1950 due to starvation, outright violence, and other indirect causes amounts to 1.2 million. However, some Tibetologists place the figure closer to half a million. The non-governmental International Commission of Jurists (ICJ) released a report to the United Nations in 1959, and another in 1960, which documented accounts of atrocities and accused the Chinese of religious-based genocide in Tibet. "
Ehh this is as biased as it probably gets and it doesn't really make it sound like genocide. I guess I had heard of this (and maybe Tibet too) as cultural genocides before, and got it mixed up.
not really, he managed to invoke nazis without it actually being godwins law. pretty impressive.
additionally, this is not a ridiculous comparison:
a) monopolistic, rights abusing dictatorship
b) large us corp ignoring this to do business there and assist with 'dual use' tools which could be used for malice against its own citizens.
> You can't just say that a government is Nazi just because you hate that government.
similarly: you can't just dismiss a government's ills or to derail an argument because you like the people from that country or the stated intent of the action being peformed there by a company you like
Well a fun trivia, a research division of Chinese Policy Department(team named Trimps-Soushen), actually won top3 positions in ImageNet competition.
I think they definitely have, at least, the engineering strength to implement what crazy, SkyNet inspired system themselves, without help from Google/DeepMind ... Which is ... pretty sad, on many levels.
All capable governments are spying on their dissidents. However, the CCP also detains dissidents on the basis of its spying even in low-profile cases[1], and has been using prisoners (in other words, more dissidents) as a source of organ "donations" for a long time[2]. All the articles I've read about this topic are about specific religious groups (Uyghurs/Falun Gong). The Chinese treatment of Islam is so openly hostile that it would make Breitbart proud[3]. China is (thankfully) not ethnically cleansing Tibetans, but they are at least cleansing its culture[4]. Also, hi from Taiwan, so thousands of CCP missiles are pointing at the city around me.
I agree that you can't compare the NSDAP to the CCP. But doing business with China is still a very different thing from dealing with most other countries (regardless of whose side they were on in 1945).
Would you say providing public education and moving people from dangerous area to safer places as a way of cleansing their cultures? If you are really interested in China, you should just go somewhere and see for yourself. Media from China and anywhere else in the world are just slanted mirrors. When some Uyghur terrorists hit a train station in southern China and killed innocent civilians, some western media titled them as "terrorist" and implied they were repressed. As for Taiwan, It's still a legacy problem of Chinese civil war, It's really unlikely that China mainland will really do anything about Taiwan, speak of missiles, actually there was a missile missent and crashed by Taiwan towards mainland in recent years.
> Would you say providing public education and moving people from dangerous area to safer places as a way of cleansing their cultures?
I actually agree that the cases of Tibetans and Uyghurs are not as black-and-white as the Western media portraits them, and I'm not sure if I should've brought up Larung Gar. But even your best-case interpretation sounds a lot like the standard justification of colonialism.
> If you are really interested in China, you should just go somewhere and see for yourself.
Been there twice. The only on-topic anecdote I have is that I went to see the Forbidden City, and a protester undressed and tried to hold up a banner at the same time. The Chinese handling was unlike anything I've seen in other countries: plain-clothed security guards wrapped her up in just seconds and carried her away (I couldn't even decipher her banner), and none of the domestic tourists around us even batted an eye. It was like a glitch in the matrix that everyone collectively pretended didn't happen. Pretty chilling, makes you wonder how many issues go unheard.
I'm planning to travel further West next time, but what do you suggest? Should I just stroll into Xinjiang and ask some locals if there's any truth to this organ harvest rumor, then come back to post on HN? That's the job of journalists, and they can't do their job.
> As for Taiwan, It's still a legacy problem of Chinese civil war
That doesn't really explain why the number of missiles keeps increasing.[1][2]
> actually there was a missile missent and crashed by Taiwan
Clumsy, sure, but we all know why Taiwan needs to handle these missiles in the first place, and it's not for offense.
The fact is Communist China has a very deep history of "cleansing" what they define as undesirables, which continues today in various forms. Today (and for decades now) they're aggressively attempting to "cleanse" homosexuals using state resources, torturing thousands of people. [1][2]
China's Communist system also "cleansed" a vast number of people throughout its recent history. Over two decades they murdered (or sent to gulags) hundreds of thousands of intellectuals and similar, who were declared to be class enemies. This in fact likely set China's potential back by decades as Mao's horrific policies severely depleted China's resources.
[1] "China urged to stop using electroshock therapy to ‘convert’ LGBT people"
So assuming what you described is true, why hasn't the rest of the world declared war against the communist China?
Because it's not close to what Nazis did. And that's my point.
Btw, from the articles that you linked, it is quite clear that treating LGBT as disease is not a government policy. In fact, it is against the law. The government doesn't actively hunt for LGBT people and send them for treatment. In fact I think they are less concerned about homosexuality than the churches are.
Your logic is not sound. North Korea was recently described as a worse human rights abuser than Nazi Germany, by a Holocaust survivor no less [0]. In spite of this commonly-held perception, no country has declared war on them, for many of the same economic and military reasons that no country has or would declare war on China.
[1] China is not similar to Nazi because the extent of "cleansing" is not on the same scale as Nazi (even if you count death sentences on illegal cult members)
[2] China is not similar to Nazi because it does not initiate new wars against other countries (Tibet was incorporated as a result of ongoing conflicts between China and other countries for decades, so it is not really an invasion)
I shouldn't have concluded that countries are not declaring war against China is due to [1]. However, it can be explained by [2]. Regardless, I believe both [1] and [2] stands.
I think you've taken more from the original comment than was given. Only you have compared Communist China to Nazi Germany on the basis of foreign policy, so that line of argument is irrelevant.
Historians criticize IBM for helping the Nazis to identify and exterminate political and ethnic minorities, not for improving their ability to wage war. In that respect, Western companies like Yahoo and Cisco have already been caught in similar arrangements with the Chinese Communist Party [1][2]. Hopefully Google's AI center will not continue this disturbing trend.
As to the scale of the CCP's "cleansing" to date, it's hard to know - they don't release any numbers, and foreign journalists are aggressively bullied, arrested, and deported if they try to find out [3][4].
We do know that China regularly humiliates and cracks down on members of its Muslim minorities [5], and that the Communist Party regularly intervenes in the appointment of Buddhist and Catholic religious leaders [6][7].
If that is what you believe, I would encourage you to check out these two books on the subject [0][1]. Both are engrossing reads, and paint a damning picture of the regime from different perspectives.
I have taken a strong amateur interest in North Korea for years, and the full body of press, reporting, and firsthand accounts I have read would support the comparison to Nazi Germany.
The world doesn't declare war on countries because of their internal wrong doings. Do you actually think ww2 was fought because the world wanted to stop German mistreatment of its undesirables?
Mao wasn’t really, he was content to let the KMT fight off the Japanese while the communists would conserve their strength for the oncoming civil war; Chiangkai Shek knew this and tried to play the same game, but the Americans wouldn’t let him.
Mao is actually on record for thanking Japan in one of his weird off the cuff remarks:
> "(Japan) doesn't have to say sorry, you had contributed towards China, why? Because had Imperial Japan did not start the war of invasion, how could we communist became mighty powerful? How could we stage the coup d'état? How could we defeat Chiang Kai Shek? How are we going to pay back you guys? No, we do not want your war reparations!"
It was more like irony, Mao also thanked Chiang who killed his some of his family members and he thanked America who killed his son. If you read history, Mao insisted on fighting against the Japanese invaders before Chang did. Do not forget KMT had America assistance in civil war, but they lost, because they were just too messy with economies and care too little about 90% of Chinese poor people.
Mao insisted on the KMT fighting Japanese invaders before Chiang did for sure, but he still didn’t bother committing substantial forces to the war. Mao also didn’t care much about poor people, see how many Chinese civilians died in the Seige of Changchun.
Mao was basically a sociopath even if Chiang wasn’t much better; he was basically a smarter Donald Trump of his time, who knows what he would have written on weibo if it existed at the time!
There was a group of communist soldiers fighting against Japanese invaders in northeast China Manchuria when KMT didn't officially do anything about that. Actually Chiang attacked communist army even when they were fighting together against Japan in later years. When it comes to Seige of Changchun, why KMT didn't go out fighting and just stay and rob people's food? Why didn't they just surrender or went out start to fight? Why waited to surrender in the end but with poor people had nothing to eat. That was a shame indeed. It was a cruel civil war, I cannot say which side is strictly morally superior, given innocent people got starved . Mao was a complicated person, there are people who thank him throughout life and people who hate him till death. He has done good and bad, he got rid of land lords, get the country united again, push hard on public education and extreme equality, but he also conducted the great leap and destructive political battles. it'll wait for probably another century that some historians can have some better estimation.
Okay, that was a rather weak point, I agree. Regardless, I do think there are more differences than similarities between the current China and the then Nazi.
Sadly no, in 20 years the whiny relativistic people in charge of whichever authoritarian regime they have legitimized via their own apathy will be complaining that people didn't think that this was a 'cool' effort, much as they are doing here.
Did anyone else here grow up being told that the Chinese government stole search technology from Google when they tried to do business there, and this is how Baidu got started?
No, that's an abnormal and rare accusation in fact.
Robin Li was working in the US on search before Google existed. That includes at IDD and Infoseek.
Circa 1999-2000, if you understood how PageRank worked (along with some other similar approaches), combined with some funding, talented engineers, and Li's background in search, it would not have been necessary to steal technology from Google to found something like Baidu.
Once www.people.cn (the website of People's Daily) built a search engine and failed. If Chinese government stole search technology from Google, why would it give it to Baidu not to a search engine set up by its mouthpiece?
Baidu had already been popular in China before Google officially entered China, so there is no 'this is how Baidu got started' thing.
No idea if they stole anything and I wouldn't/couldn't/won't make any claims about that etc. but what I've seen of Baidu's open sourced offerings, they are based on Google's publically promoted/published designs down to the types of components (their own implementations of protobuf RPC, GFS, etc.), class names, API C++ style, etc.
Lots of others have done this (some parts of Facebook, too) and I don't think this was based on Google source code or anything, but it doesn't exactly smell terribly innovative on its own?
Since the beginning, Google has claimed it was an AI company. They have killed their mantra "do no evil" and now they are putting labs in a country with lower values of ethic. Recently they were selling their robotic activity because it was too frightening. All their actions are consistent with a single aim: build an AI like in singularity or terminator.
Maybe you should not use anything Made in China, 'a country with lower values of ethic', because anything Made in China you consumed contribute to the economy of the China.
If you continue to use made in China, your criticizing Google setting up an AI lab in China is just hypocrisy.
I did not expect my comment to be interpreted as a critic of China. I have nothing against Chinese people and I am glad that Google setting in China will help the country to develop.
If Google accomplishes anything worthwhile at this research center, won't China just shut it down and re-employ those researchers at a domestic company? Why do high-tech companies even continue to try to break into China?