China’s AI Awakening(technologyreview.com) |
China’s AI Awakening(technologyreview.com) |
An objective measure is PISA results [1]. When comparing with even the best performing US state, Massachusetts, China has many more top performers in Math, as a proportion of population [2].
(In 2015 only four provinces of China participated, but their combined population was 230 million vs Massachusetts's 6.8 million. The math result of Shanghai (24 million pop.) alone would show an even larger gap.)
Since PISA results are scaled such that OECD country's mean is 500 and standard deviation is 100, China's 531 math score implies country mean at 0.3 SD above PISA mean, and US' math score at 470 implies 0.3 SD below mean. If people capable of doing AI research or proper AI implementation need to have math skills at, say, 2 SD above PISA mean, then there will be a tremendous difference in proportion between two populations with 0.6 SD difference.
My back-of-the-envelope calculation, assuming above figures, is the proportion will be about five times as large. But China has more than 4 times the population of the US, so the difference in potential numbers of AI-capable natives could be around 20 times. (Since other provinces may drag down China's mean, it could be a bit less. An opposite influence is increasing wealth and thus more resources devoted to education, both by parents and the state. We'll see soon since China as a whole will participate in PISA 2018.)
Moreover, the Chinese government is pushing AI and programming education into their education system, starting from the primary level [3].
The distinct advantage of the US is that it is the magnet for top talents and ambitious people from all over the world, including the Chinese, so it might continue to lead for a while. However, if the US becomes inhospitable to talents or potential talents, it is clear that China with its quality and quantity of human resources will soon take the lead.
[1] https://en.wikipedia.org/wiki/Programme_for_International_St...
[2] http://www.compareyourcountry.org/pisa/country/usa1?lg=en
China's advances in AI are impressive as well. A ranking based on publications at top conferences from 2012 to present (2017) has both Beijing and Tsinghua University in the world's top 5.
http://csrankings.org/#/index?ai&vision&mlmining&nlp
(The starting year in the link needs to be changed to 2012.)
Many of the top researchers in US academia are mainland Chinese. I wonder if they will be enticed back home when the Chinese government has managed to reduce the pollution in their major cities down to healthy levels. It appears that the salary for top people there, especially those with experience at major Western institutions, is comparable to those in the US in nominal terms, and the research funding is easier to get.
* We knew about Tencent's AI Lab (Saying you need your own capabilities is another way of saying: We have NIH and we will waste 3 years building our own framework from the bottom up).
* We knew about China's AI plan, which was announced in July. That was huge news at the time.
* We knew about Baidu, but Baidu is actually slipping. They lost Andrew Ng and Adam Coates, and frankly they have executed poorly on social media and cloud computing compared to Tencent and Ali. It looks like they'll lose their lead in AI.
And the article actually contradicts its own sub-headline: "The West shouldn’t fear China’s artificial-intelligence revolution. It should copy it." The West is leading in AI research, and China is copying the West. So what does it mean that the West should copy China? China is playing catch up in AI just like it has been playing catch up in GDP growth. Catching up always moves faster than blazing a trail. The West is blazing a trail now. They don't have much to learn from players trying to catch up.
I don't think you're addressing OPs point at all. They pointed out that the US currently does have the lead in the field due to its larger prosperity for the most part, but that China is rapidly developing more talent.
What we should copy is thus not the technical status quo, but our attitude towards education and talent acquisition in the relevant fields.Technical advantages are not self-perpetuating. If the West grows complacent and does not develop the talent necessary to advance the research, then that lead will reverse.
I have come to think that the country barely has a government at all. We're somehow coping amidst the chaos.
now if you are talking statistics, in particular applied statistics, most of those departments are retooling by hiring ml folks from csee departments. but the bread and butter courses required to get a stat phd remain the same as they were a decade ago - 2 core inference courses, 2 core math-stat measure theoretic courses, 2 courses on experimental stats. literally no stat dept mandates an ML course, though a few do allow ml as an elective. now, all of this can and will change over the next few years, but its very early days.
essentially, predicating ai dominance on raw math skills is a mistake. there’s no significant correlation.
You could make an argument that high performers on the PISA test would find AI and accountancy too boring and would go into pure mathematics instead; but I don't think those proportions would be different between countries, so a higher potential (schoolchildren with good math skills) should still translate into more AI researchers, on average.
For simple, routine applications, one can call machine learning APIs without deep understanding of the math behind them. In most real world applications, however, input data tend to be messy and there are many complications and constraints to satisfy. For the foreseeable future, applying machine learning in the real world requires input from humans who understand the why's and the how's behind an API.
For more advanced work, reading papers is necessary and most of the recent machine learning papers contain substantial math (from the viewpoint of an average person, not a professional mathematician).
In addition, math skills, at the basic level that PISA measures, are highly correlated with logical thinking, which will be relevant for AI engineering for a long time to come.
[1] There are perhaps fewer than 100,000 professional mathematicians in the US out of over 100 million knowledge workers there. https://mathoverflow.net/questions/5485/how-many-mathematici...
[2] https://www.noodle.com/articles/the-problem-with-college-cal...
[3] http://hechingerreport.org/high-failure-rates-spur-universit...
And for K-12 education, which is necessary for preparing students for higher education. What proportion of Americans go to schools which give them little chance at high-tech jobs?
Chairman Yang? Nah, y'all; it's gonna be Chairman Jinping.
I can only see the trend speeding up. Especially as/if the environment (air quality, etc.) improves. This now, in my personal view, seems to be the main factor for many people to decide to come or not.
> “When the Chinese government announces a plan like this, it has significant implications for the country and the economy,” says Andrew Ng, a prominent AI expert who previously oversaw AI technology and strategy at China’s biggest online search company, Baidu. “It’s a very strong signal to everyone that things will happen.”
ie. There is going to be a tonne of money poured into this.
Will it make a difference, tangibly to economic transformation? Who knows...
...but anyone sitting here scoffing that China can’t make itself a world leader in a field by simply pouring money and support onto the problem needs a serious history lesson.
They have before, and there is every indiction they will again, here with this.
More interesting: will it be as transformative as everyone is hoping?
Then, too, an East Asian country (Japan) known for excellence in one field (manufacturing) initiated a state-funded research project into a promising technology (AI). Then, too, the West reacted with alarm, fearing the Asians would leave them in the dust. Then, too, the project was about AI.
Except, back then, it didn't go too well [1]. But, perhaps this is just pattern matching failure on my part and nothing is even remotely the same today.
____________________
Chinese citizen also seems less concerned about their privacy. I've talked to one of the largest chat platforms in China, and they seems very open to share their conversation datasets with third-parties for dialogue research.
In high-value verticals like healthcare, there is also less regulation compared to U.S. For example, it's easy to start a AI + radiology company in China and have access to hospital patient data, without going through tedious approval process.
Europe and at americas combined have a larger population than china and a geographical landmass 3X larger than china.
The real question is why europe ( 800 million people ) and latin america ceded their tech industry and data to the US.
If data is the oil of the 21st century, then the US, china and india are going to rule the nest. Europe has no future and neither does south america.
I under why china, russia, etc had protected their own internal industries to compete against FB, GOOGL, TWTR, AMZN, etc. I just don't understand why europe, south america, etc haven't.
All europe/south america have/had to do is invest a few billion to capture a few trillion in wealth. Instead, they've ceded all that wealth to predominately american tech companies. It's very strange when you think about it.
I don't think Europe could do what China and Russia did even if they tried.
What's really interesting is the continued trend of cashless WeChat Pay adoption in China versus say trickling Digital Payments penetration here in USA. I actually just had to order a new replacement set of paper checks. And its simply due to the fact that so many small merchants cite higher credit card fees as a reason for preferring cash or checks for payments.
Coupled with AI, I think this preference for instant, fee-less digital payments will translate into convenience stores without cashiers. Auto repair shops where you just drop off your faulty electric vehicle and hop into a loaner without actually interacting with a customer service agent. And thousands of other technology-mediated retail experiences.
It's Asian consumer adoption and demand driving AI solutions from the bottom up. And that makes it much more likely to result in first mover advantage and the potential for game-changing breakthroughs.
In contrast, up until a few years ago, I had to bring a backpack full of 100 RMB notes to pay my rent every 3 months. China has made a transition for sure, but don’t forget the local problems they were solving (poor penetration of UnionPay, poor online payment facilities). And online retail and a godsend when you consider the horrible state owned department store system and anemic brick and mortar retail clones they replaced.
Chinese companies will probably change the game in china, it as long as they are focused and limited to local problems, I don’t see how it will effect the west very much.
Despite China's outward enthusiasm for advancement, high speed change promises to destabilize the totalitarian gov't, which suggests that a bumpy ride lies ahead.
First, even though Chinese high schoolers are "better" at math and there are certainly many gifted kids, it does not necessarily mean that they are better at abstract thinking and will excel at higher levels. On the other hand, some of the most talented people I know did not have super impressive test scores. They are insatiably curious and think from first principles.
Second, if you look at the top researchers today, you'll find their strength comes from combining knowledge from different disciplines, e.g. computer science and neuroscience. The education system in China pretty much hones students in a single dimension, pushing them into a narrow corner. Also, the most important thing in research is not just understanding what has been done. It is thinking critically and challenging authority. Unfortunately, in my opinion, this really goes against the Chinese regime and Chinese culture to a large extent.
I wouldn't be surprised by a large number of future Chinese AI "engineers" who can quickly replicate the latest research. But I think China has a long way to go in order to create an environment that allows students to freely explore and think before a real AI breakthrough.
China's push for AI is a recognition that the loss of one advantage (cheap human labor) could be replaced by another (cheap machine labor).
In any case, the advantage of cheap human labour has already been replaced by what you call 'cheap machine labour'. China has been able to dominate manufacturing in large part by giving companies access to cheap capital, which allows them to invest in machines. Silicon chips aren't made by hand. PCBs aren't soldered by hand.
/sarcasm
Really? Major breakthroughs can be mandated by a centralized government? Back in the 80s-90s, Americans were convinced that Japan was going to eat its lunch. It gives me some hope in knowing that we still suffer from this complex.
Computers, nukes, ICs, space, internet, self-driving cars, etc.
They seem to be some of the most well used research money -- precisely the government mandating progress.
Based on their keen interest in AI, the Chinese gov't presumably doesn't yet appreciate that success in task-specific AI does not assure the advent of AGI any time soon. I suspect like the Japanese's 5th gen AI initiative, China's quest to give life to terra cotta warriors also will end in disappointment, before the passage of several 5-year plans, anyway.
But I'm looking for things where the technology has yet to be applied.
There is also logistics (again tons of messy bits), customer service (interacting with people), transportation (now this is probably the lowest hanging fruit - which is why everyone and their dog is chasing it).
Just of the top of my hat.. of course there is R&D (hard), design (hard), bureaucracy (bookkeeping, process management, etc. - easy in theory, not so much in practice) -- well, and all the other things that we humans do during a day of business.
We could also involve politics, government and philosophy - but that would probably be stretching it.
ps. I also suck at counting, so maybe that too.
top 20 list:
https://www.usnews.com/education/best-global-universities/se...
https://www.usnews.com/education/best-global-universities/co...
According to Wiki, three more independent and open source of rankings are the Acad. Ranking of World Univs, Times Higher Education, and Quaccquarelli Symonds.
https://en.wikipedia.org/wiki/College_and_university_ranking...
The Chronicle of Higher Education is also a well-regarded rater, esp. in assessment of research.
http://www.chronicle.com/article/The-World-University-Rankin...
Also, not surprisingly, most quants were Chinese.
Aren’t a ton of quants French as well?
AI is the new moon shot of our time but its rally cries are emanating by leaders of U.S corporations for the maximization of profit of their shareholders where as in China they are emanating by the leaders of the state for reasons far exceeding short-term economic gains
What are they currently leading?
China is leading in pretty much everything right now. They are the top trading partner of pretty much every significant country. And they are planning on building another 100 cities in the next 10 years.
It is ridiculous how much change china is going through right now. Historically, urbanization/cities correlated with economic/military strength. Will that trend hold.
China currently has 100 million man cities...
https://www.theguardian.com/cities/2017/mar/20/china-100-cit...
They are planning to double that in the next 10 years.
We have 10 cities that are 1 million+ people. Europe ( including turkey and russia ) has 33 million man cities.
Think about that.
Of course there are a lot of political factors in both of those achievements. But money definitely translated into results.
(I'm not arguing that China is the best nation in this area either, but they are clearly a leader.)
Look at their debt and then tell me they dont try.
Certainly that example helped advance genomic science and the biotech industry in America. But it didn't quite live up to the hype, and it hasn't exactly saved the US economy from the normal ups and downs of global competition.
Course ya never know, could be the start of the next ARPANET too.
My point is that trajectory is more relevant to these discussions than current state of affairs. The quality of life of the average Chinese citizen (and more so the upper-middle class Chinese citizen) has increased DRAMATICALLY in the last 40 years. So when people talk about China's potential, they aren't talking about the situation in the next 5 or 10 years. They're talking about who is going exert massive international influence in the next 50 years. So you're right, most westerners aren't buying speculative property in Beijing (despite the fact that housing in central Beijing is nearly as expensive as New York City), but I'd wager there's a good chance of that changing in the next 50 years.
Western countries have a much better record in this regard.
Also, even if housing forsake prices in Beijing are at New York levels, rents are just at Wichita levels. No foreigner is going to buy into that level of crazy, especially when buying property in china accords them no residency benefits (they will never get hukou, heck, green cards basically don’t exist).
Why are rich parents sending their kids to the US? There's a shortage of jobs for college graduates in China and many parents are hoping that a degree from a western college will give them a better chance at finding a job either in the US or China.
As for buying property. The biggest reason is probably diversification and the fact that the Chinese see property as a good investment. Housing prices in China have sky rocketed and many Chinese families have more than one house that they just leave around with no one living in them. Just like how stock investors invest in foreign stocks, Chinese landowners are diversifying their property.
Foreign investors don't want to buy property in China probably because of the pollution problem and the fact that they think that the housing market there is a huge bubble.
We're not interested in ideological battle or flamewars here, and certainly not nationalistic ones.
Edit: since you've done this repeatedly, I've banned this account.
I, however, believe that whoever wins this battle will be wealthy and powerful. Who-ever wins the race is the enemy. Not just the Chinese, but anyone who attains singularity is the enemy of the rest of humanity.
Even if that AI is open-source, unless there is a clear way for every person to have some level of control over the hardware that it runs on, the AI becomes the enemy of all people who will be disenfranchised by it.
Because, so far in human history, individual rights-based political formulations are the only ones that prevent arbitrary atrocities from being perpetrated on citizens. Yes, American corporatism is brutal. American corporatism bounded by American democracy, imperfect as it is, is still vast leagues better (read: safer for global population) than nationalist authoritarian dictatorship. No matter how technocratically efficient that dictatorship has been in recent decades.
In short, if you're worried about singularities, aren't you worried about the priorities and command structure of the entities most likely to achieve it? About what options they consider to be "on the table"?
For example I was randomly looking through text recognition competition results and found out Tencent has an AI lab that took first place: http://rrc.cvc.uab.es/?ch=1&com=evaluation&task=4.
But I was unable to find any information about them. They have github repos with thousands of stars but the community is entirely in chinese: https://github.com/Tencent/ncnn/issues
Like, social websites and apps?
>What breakthroughs have come from China?
Like tons of actual innovations in hardware, screens, storage, materials science, manufacturing methods, robotics, biology, and other fields?
That was actually developed in Silicon Valley though, as that's where Baidu's AI lab is.
https://www.forbes.com/sites/jpmorganchase/2017/10/04/chicag...
DeepMind is at Google which is a Silicon Valley co.
True about Canada, but salaries in Canada do not support retention of talent, so most end up in SV in the long run or get acquired by SV companies.
US research institutes are top-notch, but our casual racism and xenophobia are killing us. Not that the Chinese aren't also racist and xenophobic, but the pendulum of power is definitely swinging back across the Pacific.
And I'm not sure Russia can be considered a locked down regime in recent years. Unless you mean specifically Putin's grasp on power.
I don't think transportation is very easy, it's just that the value is high and honestly, there is this trillion dollar asset (the nation's roadways) that companies are going to be able to leverage for a very low cost, which makes it appealing.
Why do you believe that?
To answer your question, I believe that because I haven't yet seen a Deep Learning paper which did not couch its results in terms of those, even if it might not have been strictly necessary. If you do not understand the basics of current approaches, you'll have a hard time developing them further.
And many failures too. You can't just buy whatever innovation you want with money; if that were true we wouldn't still have energy problems and powerlines for distributing it.
I’m skeptical that’s our current political climate in the US would be willing to do something like that again unless there becomes an “AI race" (which isn't far fetched, but not guaranteed either).
I bet China eventually scraps the lease system and replaces it with a universal property tax, which would solve the other problem of keeping local governments funded long term (right now they just make money on sales), while naturally deprecating the property on a yearly basis, and also making pure speculation much more difficult.
But who can say. If china switches to a property tax system (not just the nominal one easily gamed in shanghai and chongqing today), it would break a lot of buying assumptions.
I have some confidence of this if I purchase property in a Western jurisdiction, but in China the situation is a bit more fluid. I have zero leverage and have no doubt I would get trampled if convenient.
This is the perception they have to overcome, and to be honest, I don’t think they care to; do they really need or want my capital invested in China, especially in property?
Signals from ruling class appear to indicate otherwise.
I think pollution in China will likely be a non issue in 10-20 years or so. Pittsburg used to be like Beijing, but as the National economy shifted away from industries causing pollution, it disappeared entirely. China is in the midst of this process.
We were told in 2010 that the government would solve China’s pollution problem probably by 2015, 2020 at the latest. Ya, well, that isn’t going to come to pass.
I was under the impression that Cambodians are mostly using devices from Chinese brands and hence exposed to pre-installed Chinese software (at least to a small extent).
For example, Xiaomi phones come without google play stores but Chinese app stores.
http://geeksincambodia.com/cambodias-2017-social-media-digit...
Chinese phones outside of china are generally sold with a straight google app stack. None of those Chinese services are localized in kmher!
According to this page,number 8,9,10 ranked Android apps (from play store) are app offered by Chinese companies.
Also, there are cheap Android phones directly brought over from China which are installed with Chinese app stores (I know that's the case for Myanmar, not sure about Cambodia), they wound be missing from this chart.
So I wouldn't say it's entirely MIA.
They are also developing production low-speed maglev trains.
Not many actual breakthroughs come from China so far, even though they could fill a few Olympic pools with PhDs.
Many young Chinese who were educated in the US feel this way, so it's not a generational thing. This is a cultural tradition stretching back many thousands of years. The world view of a person from China is very different than that of an American.
Also, consider that an AI researcher returning to China is likely going to be in the top 1% of society. Dictatorial societies tend to be pretty good for the people at the top. The Chinese startup market is hotter than the US at this point too; and it's basically restricted to people born in a handful of Chinese cities. The Chinese government intentionally stacks the deck to get them to return.
All things considered, the US looks really bad right now internationally. We have vast military power, and our leader is a child elected on a wave of nationalist populism. This has happened before in recent history, and it didn't end well.
It seems in China excellence and competition are truly appreciated, from family, school all the way to the government, even though some of them have to use VPN to use Google.
China sets a national goal each five years, education first has always be the first priority for the society, back in USA, we're debating how to borrow more debt, how to make sure people can choose no-gender(neither male or female), how to remove ranking from school because competition there hurt our 18-year-old-KID's self-esteem,etc, what a joke, we're doomed.
We've asked you repeatedly not to do things like this, and your comment history doesn't exactly show you using the site as intended. That's not cool. Please read https://news.ycombinator.com/newsguidelines.html and fix this if you want to keep commenting here.
I expect World War 3 will eventually be fought over these differences; they are just too irreconcilable and eventually the Chinese are going to feel the need to impose some limits on the US -- likely over environmental issues. China will feel the need to "save the planet from US ignorance" once they feel they have the power to do so, and the US will disagree that they have the power, so China will have to prove it militarily.
And that's funny because if there is something that the US is not it is a democracy. It is at best a partial democracy where votes from different individuals count very different towards the outcome and where there is a huge and concerted effort at the repression of certain groups of voters.
Americans tend to be insular. A very low percentage of Americans have a passport.
The healthcare debate is actually a good example. It’s entirely uninformed by how systems work elsewhere. As a foreigner it’s baffling to watch.
I don't know if people have forgot about them, but I am pretty sure many - maybe most - have forgotten (or just don't want to contemplate) what kind of damage they can unleash, both immediately and over the long term - should a full global exchange occur.
Only if war from general misery due to overpopulation and ecological collapse doesn't get us first. I hate thinking about the future.
For example, a medication that has a 70% success rate can be seen as risky or not risky, depending on the person.
If anything, the modern era should be a boon to them, since developed countries are virtually conquerable at a certain scale, via modern technology instantly turning that industrial base into a top tier military, if needed, and via the usually strong alliances provided by strong trade relations.
I actually believe China can be a peaceful #1, provided they can be stopped from bullying their neighbors and Africa when the latest political scandal demands a distraction (à la Clinton - Kosovo).
Is extrapolating from the past really useful if the Chinese government has been completely replaced in a civil war, which was followed by the massive changes of the cultural revolution?
Likewise, is Germany a dangerous country right now because "we" have started 100% of world wars within the last century? Honestly, I wouldn't even trust our military to capture Luxembourg right now... (probably for the better!)
This applies to all aspects of China's development, of course, not just their military.
Perhaps they were not, though they were so dominant that they had no security fears and had other levers of power.
Contemporary China has militarily threatened or even had conflicts with most of their neighbors, including Japan, The Philippines, Vietnam, India, and Taiwan, and also the U.S. Communist China fought a war with Russia in the mid-20th century, invaded S. Korea (with the N. Koreans' assent), and invaded and annexed Tibet. They have built-military and quasi- military bases in Sri Lanka (not sure of the status of that one), Pakistan, and Djibouti. They've built infrastructure supporting potential military logistics throughout Asia.
The Chinese government's talking point is in the comment above: China does not interfere in other countries affairs and they seek a 'peaceful rise', citing their history as if President Xi restricts himself to what was done in the Qing, Ming or prior dynasties centuries ago. Almost all warmongers in history, or the ones with any sense, have made the same claims about peace; look up some German propaganda from the 1930s.
0: https://www.berkeley.edu/news/media/releases/2003/10/27_lako...
sorry Trump, it is too late to fix, the tipping point has passed.
I wouldn't call 50% state ownership of most if not all large companies anything but "socialist"; they just found a way to introduce incentives into a socialist system by abandoning the idea that everyone should be a "worker". But wealth redistribution is absolutely at the center of the Chinese system.