Many simply don't know. For instance, Liu Xiaobo was banned from public speaking and it was banned to publish his writing. I won't be surprised if he was better known outside China than inside.
No idea why this is being downvoted. It's a very salient point. The death of a nobel laureate while in captivity as a political prisoner last occurred in Germany in 1938. Much as people should have been scared by German rocket and tank tech in the 30's, they should be scared of Chinese AI tech today. its not an unreasonable assumption that this technology is going to be used for things that run counter to the values of many people in US, with the enhancement of the domestic security apparatus and control of the media chief among them.
The article is subtitled "Its deep pool of data may let it lead in artificial intelligence" for a reason, you can't just pull a "well trained" model out of thin air, and proprietary data is absolutely defensible.
You'd think China would have an advantage here. But the kind of special data only Google and FB have is mostly related to ads and people profiles. That's not the kind of data you need to advance AI.
Instead, what is needed is user labeled images, videos, text, speech and of course, simulations (games, VR) for virtual robotics and reinforcement learning agents. These kinds of data and simulations are open sourced and growing. Once there is a sufficient collection of training data, it's cheap to train your neural nets.
Even today, there is much more data than scientists care to use. For example, ImageNet is over 1TB, but a smaller portion of it is used routinely (it has 21841 topics but usually just 1000 of them are used in research papers). So it's not the lack of images that's keeping them. China would have no advantage here. Everyone has approximately the same level, the best paper of last year is the average of this year. That's about AI in general - but regarding self driving cars and ads Google is cooking in secret so it might be a little more ahead.
> But the kind of special data only Google and FB have is mostly related to ads and people profiles. That's not the kind of data you need to advance AI.
> Instead, what is needed is user labeled images, videos, text, speech.
umm:
google image search, all html embedded/related metadata, adwords itself serving as a refined metadata/tagging and ontology database building engine, recaptcha 'how many buildings are in this picture' puzzles, android debug settings where things are uploaded, google drive cloud storage, gmail, the google speaker gizmo and voice activated phone agent, google voice voip, google chat, maps annotations to pictures/gps coordinates, etc etc etc and yadda yadda yadda...
not sure what else you need as far as those sources..
Everyone has access to image search results (from multiple providers). Ontologies such as DBpedia and WikiData with billions of triples are open source, don't see why they couldn't closely track closed-doors ontologies such as the one used by Google. The picture puzzles are useless - we already have superhuman ability to classify traffic signs and storefronts. Android debug data and map data are irrelevant for basic AI research.
They've got a lot more people, so it's just a matter of time. That, and there aren't any real "trade secrets" when it comes to AI. It's all just computer code, and anyone with an Internet connection can teach themselves how to use Wikipedia and run TensorFlow.
Sure, you can go through some tutorials and learn to build a CNN, or even read more recent research papers and hit bleeding-edge ImageNet benchmarks, but nothing currently public is going to get you close to Google's SDC.
You can't pretend this is just about people either, India's at 1.3B and they're far from becoming a world leader.
This really depends on what is the driver of the current rise of neural networks and other similar machine learning techniques... whats the contributions of data and method toward the new improvements in performance? If more data is main contribution I would venture to say that China can indeed beat USA.
It's a combination of recent innovations and more data. Having access to more annotated/labeled data has definitely propelled the state of the art in some disciplines like machine vision.
Wait, what does Wikipedia have to do with TensorFlow? I think you are over-simplifying AI. Yes, you can take pre-trained models are do useful things with them. But creating these models and training them initially on vast quantities of data still remains a challenge and has to be overcome by real engineers.
For that you'd have to wait until next year. Many low level skills necessary in ML are being automated. It's getting closer to being easily used by hackers.
TensorFlow is the trendiest ML library, and Wikipedia is a source of knowledge. You can gain the knowledge required to effectively use TensorFlow (or whatever) from freely available sources, provided you have the interest and tenacity.
Chinese Government is all over the Silicon Valley these days trying to get into US AI developments, using all sorts of nefarious tricks.
A case in point. There is a conference, SYNC 2017, in Computer History Museum tomorrow (July 14th), organized by Pingwest, which itself seems like a Chinese Gov. front if you look at the team.
They have several good speakers, and some Chinese, Gov fronted, fake VC types.
Till 2 days ago, it was titled SYNC 2017 - Unbounded: Building Across Borders.
Yesterday, it's title was changed to SYNC 2017 - US China AI Summit
It is pretty clear that if they had gone with the US China AI Summit in the beginning several of their speakers may not have agreed, but now it is too late for anyone to change mind.
Edited: To change from "Chinese" to "Chinese Gov." to clarify I'm referring to the Chinese Gov and not Chinese people
You described Chinese people organizing and attending an open technical conference with permission, and offer the unproven statement that several speakers would have refused to speak if they had known just how "Chinese" it was going to be ahead of time.
I'm not sure how those speakers couldn't just cancel, sure they'd lose any speaking fees but I've never heard of lock-in speaking contracts for conferences. Or if they care so much they could just modify their speeches to not supply anything of value. Or perhaps they should have done their homework ahead of time.
In any case I'd hardly call it a "nefarious trick" regardless of who was doing it. I'm hesitant to throw out "racist", but you seem to be leaning that way...
If their goal was a bait-and-switch then why change the title? If you're trying to con someone out of information why make the victim aware of your intentions right before the con? It is possible that they thought "US China AI Summit" would be a little alienating so to get some initial sign-ups they put out something softer, but that just seems like a marketing decision. When I hear "Building Across Borders" my first thought is "Oh, it's an international conference. Let's see which countries will be there! Oh look it's a ton of Chinese people...", not really concealing anything beyond the superficial.
I have no love of the Chinese government, but this is a free exchange of ideas that is open to the public (for a $40 ticket). Given the actions of the Chinese government to date, if this is how they plan to get more technical data going into the future I'm hoping they do more of this and less hacking of proprietary/classified info. :P
By fake I mean fronts for the Chinese Gov. There have been numerous reports in NYTimes and other reputable media on this topic, including many that have been posted here on HN.
Of course not all of them are such, but many are. Just look at the team behind many of these, team's credentials and experience, and in many cases their addresses and actual offices.
To be fair, China is a AI powerhouse in its own right. China is investing a lot of its money domestically on this problem as well. Lots of legitimate papers come from purely Chinese teams. ResNet was developed by a purely Chinese team. A lot of very important open source machine learning software out there is written almost entirely by Chinese nationals.
It's important to note this because shutting down Silicon Valley/US access to Chinese VCs and companies won't necessarily just harm them. It would harm us as well. It does need to be looked at though, just carefully.
And by luck. If they didn't do it, someone else would have done it, whether it be US, UK, or Canada. Proof? Look at the developments and publications from that same team. There are not consistent breakthroughs.
This is similar to the stock market and investors. Someone will get lucky and win, but unless they can do it consistently, it's luck.
There are similar problems on both sides of any nationalistic argument, and both sides routinely violate the rules here, as indeed you have been doing here and elsewhere. Would you please stop doing that?
I do know a decent number of top Chinese VCs looking into investments in Silicon Valley. Those funds are typically started by Chinese entrepreneurs who made a fortune in things like the internet wave or real estates.
There do exist branches of Chinese government, or nation-owned organization making investments in US. But they are not the top VCs, and I haven't seen any of them taking a stake in a top US AI startup.
There are a couple of reasons why Chinese money is flooding into silicon valley. First, there is too much cash in the domestic market and not enough good tech investments to make. Second, there is a fear that Chinese economy might crash someday so having oversea investment diversifies the risk.
Hey guys. Chen here. I am an editor at PingWest and one of the organizer of this event.
From what I know we don't have government background or funding. PingWest is a privately owned and funded media company in China. We run one of the most influential tech blogs there. Our US team organize SYNC events every year in the Valley and some pop up SYNC events elsewhere.
We changed the title simply because it fits the event and our speakers lineup better.
organized by Pingwest, which itself seems like a Chinese Gov. front if you look at the team.
I am PingWest subscriber on WeChat and generally quite fond of their writing. Last night, I privately and specifically asked Thomas Luo [1][2], the co-founder of PingWest, whether they have connection to Chinese government. Thomas denied the claim. Also looking at his resume, there isn't much I would raise the flags of being a government front. So my question for you is this: what are you smoking?
'The Yellow Peril (also Yellow Terror and Yellow Spectre) is a racist color-metaphor that is integral to the xenophobic theory of colonialism: that the peoples of East Asia are a danger to the Western world. As a psycho-cultural perception of menace from the East, fear of the Yellow Peril was more racial than national, a fear derived, not from concern with a specific source of danger, from any one country or people, but from a vaguely ominous, existential fear of the vast, faceless, nameless horde of yellow people opposite the Western world. As a form of xenophobia, the Yellow Terror is the white race's fear of the rising tide of colored people from the Orient.[1]'
It's entirely possible that you have a valid point to make, and maybe it's even an interesting and important one, but maybe also take a second to consider if it might be motivated by racist tropes.
Why do you think this has anything to do with racism? Is anything claiming an Asian country is doing better than a white country Yellow Peril? If not, why is this an instance of it?
I don't see how this article can be related to that Wikipedia. it's overly reactive and exhibited a lack of self confidence from the people felt offended.
70% of the US budget is inexorably allocated to entitlements, interest on debt, and welfare. The rubicon of 100% US Debt-to GDP will soon be crossed. Many US states are hanging on to solvency by a thread. Corporations and banks have sequestered hundreds of billions in cash waiting for signs of US fiscal stability. And what does the media want to talk about? Russian spies and Chinese A.I.
Rome burns and everyone is asleep or mad with partisan fever.
You are just fear mongering. USA's debt to GDP ratio is not a problem for our economy or anything else and won't be for a very long time. The US economy is growing fast enough that the debt-to-GDP has no effect on our stability as we could pay it off if we wanted to and do indeed have plans for this sort of thing. Also most of our debt is actually owned by US citizens which also helps make this a non issue. There's a reason nobody credible thinks the sky is falling. It's not.
Debt suppresses growth above 85% of GDP. At current levels it reduces net purchasing power. It depletes tangible foreign reserves. It causes legislative gridlock. It distorts the Federal Reserve's balance sheet.
It also doesn't sit in isolation. The BLS statistics add another 200 or so percent for about 300% debt-to-GDP when you include household, state and muni debt.
Unlike China, the US citizen lender argument isn't helpful. US debt holders are free to move their money into other assets.
The only argument in favor of the current debt levels is that no other nation can float enough liquidity to compete with US debt. For that reason alone the sky isn't falling.
Medicare and Social Security are not entitlements. Social Security is a pension system that we pay into. There are some exceptions to this for Social Security but a large majority of the funds payed out are to people who paid into the system. Medicare is an insurance system for the elderly. We all pay into it. Social Security and Medicare happen to be government run and there aren't options to opt out of them but they are not welfare or entitlement programs.
The shortcomings for Social Security are easy to fix. I don't know about Medicare and it's actuarial prospects.
Crossing the 100% debt/gdp ratio is not a Rubicon. Lots of countries have done this and been none the worse for having done so. The U.S. famously exceeded 100% debt/gdp ratio during WWII. It was able to come down from that.
Current debt obligations for the U.S. are in low interest bonds. The borrowing of the money has not crippled the nation. Spending it poorly on stupid wars, a massive surveillance/security apparatus in the form of DOD, CIA, etc. has been a problem. Currently, as a percent of GDP taxes in the U.S. are quite low.
China has 300%+ GDP to Debt ratio, collapsing economy, declining forex, fake GDP numbers, strict capital controls, internet censorship, middle income trap, authoritarian regime that brutalizes treatment of its own people, floods the world with cheap, shoddy products that destroys the manufacturing and community of other countries, pollution, poisonous food, poisonous water, mindwashing culture, allies with NK that seeks to destroy the world, pollutes the earth, etc.
And what do people want to talk about? how bad (supposedly) US is. When US leads the world in innovation and protection against aggression from NK and Russia (and China), ruled by dictators that brutalizes its own people. Whataboutism.
This sort of ideological rant isn't welcome here, and we've banned this account. Please don't create accounts to do this. It's not what the site is for.
Btw, The Economist is a British media, not an U.S. media. So your assumption of domestic media not talking about the bigger fiscal issues in US doesn't really apply in this particular case. And traditionally, The Economist has covered a lot of aspect of China in a fair and analytical angle than any other US medias do.
70% of the US federal government budget is 15% of GDP. Not such a big deal, especially considering that we're going to be paying for pensions and healthcare somehow, whether as a government program or otherwise.
Rubicon? I don't recall the sacred oath not to cross a 100% debt-to-GDP ratio. What's the importance of that particular round number, which many countries including the US have crossed in the past, except for rhetoric?
Finally, what makes you think corporations are holding cash waiting for fiscal stability, of all things? How would US dollars or bonds help them in case of US fiscal instability? Isn't it more parsimonious to think that they're waiting for return on investment? And that has much more to do with growth than with government debt.
10% goes to welfare.
6% goes to interest on debt.
16% goes to defense.
25% goes to CHIP/medicare/medicaid.
(25% to social security, but that has a separate tax inflow and is a net positive to the budget.)
The Reinhart-Rogoff research indicating that there's a public-debt-to-GDP Rubicon was wholly discredited by grad students; turns out there is no relationship between debt-to-GDP and growth.
The broader point is, if I issue money and then issue obligations on that money, there's no practical difference between the two issues. That is, public debt of countries that are sovereign in their currency, and only issue debt in their currency, cannot have public debt overhang. It's just a red herring, based on a false analogy between governments (currency issuers) and households (currency users).
Historically, rates above 200% (eg. Argentina) lead to disorderly corrections (crisis).
Michael Pettis regarding the research you cite states: "What is usually missing from such discussions, however, is an explanation of how too much debt leads to slower growth." ... "A country has too much debt when stakeholders begin behaving in ways that reduce growth. The controversy over the research by Rogoff and Reinhart should not be allowed to obscure the nature of the link between debt and economic performance."
As I said, countries that are sovereign in their currency and don't hold any foreign-denominated debt cannot suffer debt overhang. Or put another way (to specifically address your link, though I only got four paragraphs), there is no involuntary default risk if we're talking about a country issuing money and then issuing liabilities denominated in that money.
Regarding specific claims in your linked article:
- Bond markets in such countries are subordinate to central banks; the most obvious case is Japan, where a greater-than-200% debt ratio is accompanied by, I believe, negative yields on public debt. Central banks (again, in countries that have nonconvertible sovereign currency) totally control short-term interest rates, and so indirectly control long-term rates. Creditors take the central bank's number and then risk-adjust and discount it for borrowers; for the government the risk is zero, so they just discount it.
- Ricardian equivalence is a stupid model. Have you ever made consumption or investment decisions based on a belief that public deficits will result in higher tax rates at some point in the future? I certainly haven't.
- "Workers organize to protect their jobs," but I guess the anti-union bent says a lot. Nevermind that spending equals income; a falling wage share is a far more plausible explanation for less-than-desirable growth in most situations.
- "savers [...] withdraw their bank deposits" The loanable funds model is not how banking works, but that's a much lengthier discussion. Suffice it to say, it's still the case that central banks set interest rates. I assume this is the reason the author brings this up, because otherwise the question is---they withdraw their savings and do what?
- Again, the only way the US government can default on US Treasury obligations is if Congress explicitly decides to do so. There's no mechanical or functional reason this might happen.
Historically, countries have not been sovereign in their currency. You'd have to be more specific about Argentina, though its biggest debt crisis was because its currency was pegged to the US dollar. Likewise, the current troubles in the Eurozone come from a lack of currency sovereignty. These things are unfortunate, and will probably lead to a dissolution of the EMU at some point, but they have no bearing on the US.
China has an even worse problem because of their 1 child policy. An aging population with no safety nets who will have to depend on their lone child to support them.
We can always print more debt instantaneously thanks to the FED. China can't print more humans.
We have our problems for sure, but china is facing far greater problems.
Here I was looking for the Dr. Strangelove equivalent of the "AI Gap" :-)
Given that in the next four years it is likely that more than half of the PhDs awarded by top US schools are awarded to foreign nationals[1], and our current government policy is to restrict immigration as much as possible, we are on a path to insuring that soon the world will have a much better pool of talent to draw on to do the research than the US will.
China also restricts democracy and access to the internet, but that's not to say the US wouldn't be harmed by a move in that direction. China has different strengths and weaknesses; the US would be ill advised to give up its own strengths.
Not to sound like an anti-intellectual, but it could be a mistake to assume that innovation comes from academia.
Nassim Nicholas Taleb has a convincing chapter (I think it is Chapter 13: Lecturing Birds to Fly) in his book "Antifragile" about this very thing.
Then there are cultural differences that could be at play concerning innovation. It is a general observation that east asian cultures tend to suppress risk-taking and innovation. Whereas, western cultures seem to promote risk-taking and innovation. That is not to say that there aren't risk takers in China and risk averse individuals in the West...just that overall there may be asymmetry. And, very little asymmetry can create very big differences.
So, have all the smartest people you want, doing all the research you want; but if they have no pressure to put that research to practice, nor a desire or support to take the risks associated with implementation then you might as well have nobody doing anything.
Friendly writing advice: use simple words. Elongated romance (etymologically-Latin) terms sound pompous. Short words are better. Think "gut" not "stomach".
I don't think that line of reasoning is completely wrong but I think it's often over-emphasized as some sort of durable higher truth rather than a simplification of very complicated social factors, and usually this tends to significantly under-weight the ability to change things — e.g. many Americans are used to thinking that we have a huge cultural lead for entrepreneurship but the popular opinion there hasn't adjusted for the degree to which Europe, Canada, Israel, etc. have closed that gap over the last couple decades.
In the case of China, I don't think there's any question that it used to have higher barriers to innovation but I'm not comfortable concluding that this will be true for that much longer given how many people are trying to change it. Every time I read one of Bunnie Huang's gongkai posts[1] it really seems like the view in 20 years will be radically different.
I agree. I wasn't using it as de-facto; rather as an idea that may calm the panic of "OMG They have more X than we do!" "They have more X than we do" only works to determine an outcome if all other things are perfectly equal, which they are not. My primary point was the idea that academia does not necessarily mean applied innovation.
That said, how long does it take for a culture to change part of its core structure? Cultures seem to me evolutionary and, aside from rare radical mutations, very slow to change. Then again, with over a 1.3 billion people there are sure to be plenty of sub-cultures with enough size to be a driving force for the whole.
> Not to sound like an anti-intellectual, but it could be a mistake to assume that innovation comes from academia
It isn't that innovation necessarily comes from academia, but that academia is a reflection of how wealthy and educated the population is. Innovation comes from the wealthy and the educated. So as china gets wealthier, they will innovate more.
> It is a general observation that east asian cultures tend to suppress risk-taking and innovation.
That's an "observation" based on racism and nonsense.
> just that overall there may be asymmetry.
Based on what evidence? It's just silly things people say - "The west is more individualistic". A society based on "white supremacy" is "individualistic"? How is basing one's ideology on "race - a genetic/communal" concept individualistic.
All these things are essentially nonsense created by people who want to sell books/etc. There is no truth.
You could write a book saying the west is collectivistic and china is individualistic by cherrypicking any data.
> So, have all the smartest people you want, doing all the research you want; but if they have no pressure to put that research to practice, nor a desire or support to take the risks associated with implementation then you might as well have nobody doing anything.
What makes you think they won't?
Greed works on the chinese as much as westerners. The things you need to "innovate" are wealth and security. Innovation is a luxury. The chinese will innovate as they get wealth and security.
The chinese are risk averse and yet they love gambling. They are risk averse and chinese have immigrated all over the world.
> academia is a reflection of how wealthy and educated the population is
This is correct, but that doesn't mean causation. Look at any civilization and look at what came first wealth or academia. It seems to me that wealth creates academia.
> That's an "observation" based on racism and nonsense.
Saying something is "racist" does not constitute an argument. I didn't say whether it was "good" or "bad" to have a culture that promotes or suppresses risk taking...no one can. It is not "racist" to say that two things are not exactly the same.
> The things you need to "innovate" are wealth and security. Innovation is a luxury.
Dead wrong. Wealth and security create complacency. "Necessity is the mother of invention" is not just some stupid quip.
>What makes you think they won't?
I don't. My entire post was a postulation, which is why I used phrases such as "it could be", "general observation" and "there may be", etc. It appears that you are merely looking for a fight instead of having a conversation.
Yes. That's was my point. You get wealthy, then you get "civilized".
> Saying something is "racist" does not constitute an argument.
No. But saying an "observation" based on "racism" is an argument.
> I didn't say whether it was "good" or "bad" to have a culture that promotes or suppresses risk taking...no one can.
Well you did imply it...
> Dead wrong. Wealth and security create complacency.
Not dead wrong. Dead truth. It can for some people, but it also gives people FREE TIME to pursue other things. I didn't say EVERYONE would become innovative. My point is that you need wealth and security to be innovative.
>"Necessity is the mother of invention" is not just some stupid quip.
Like all quotes. It is. And without wealth and security, not much time or resources to innovate.
> I don't. My entire post was a postulation, which is why I used phrases such as "it could be", "general observation" and "there may be", etc.
You don't but that's a lot of "implying" what you think.
> It appears that you are merely looking for a fight instead of having a conversation.
I would but you seem quite defensive. I wasn't calling you racist. I said the observation was based on "racism". Lumping 1.4 billion into one "stereotype" is the definition of racism.
Also, as I noted, I debunked most of your "it could be"s.
A key Chinese collaborator of mine on this https://hackaday.io/project/21966-quamera and I have chatted about the 'old' China and the 'new' China - he thought up, designed, and built a company around making a general purpose FPGA that allows for a wide range of imagers to be attached via standardized I2C/SPI or USB bussing. He's very much the new China, and while there are cultural differences, there's no question that an ever growing number can be either incredible collaborators or fearsome competitors - a lot of that choice is up to us.
Arguably, it is better for the entire humanity if PhDs and other talent are spread out across the whole world rather than monopolized by one single nation. That's how knowledge gets transferred and the whole world improves.
I guess it's worth asking - what will the differences between Western and Eastern AI be, if any? Will/can cultural differences in philosophy make their way into competing AI via Conway's Law?
Mostly the advantage that China has, other than probably a large pool of math educated engineers and abundant data thanks to China's aggressive surveillance, is that things would be unethical or hard to move forwarded under current regulation, would be much easier in China.
Ironically, underlining in this article is that acknowledgement of such often unethical or act in doubt are perhaps essential and critical to AI's future development. To that, China is at a clear advantage than America, and the rest of civilized west world.
A strong economy is also necessary to propel the development, or Russia would also be at a advantage similar to China's, or North Karea's. But only China has the economy incentivization similar to West's.
first, it catches up with economics, second, the rising middle class demand for more democracy to upgrade the existing political system, smoothly hopefully, and last they begin to beat US all around, purely due to the population and an education first tradition, with east-asia's above-average IQ as the base and the much-better economy as the booster.
Well, China both educates its own people instead of foreign national freeloaders ... and it sends Chinese people to the USA in order to pick up on AI and other technologies ... and they have been digitally stealing, robbing, and pilfering everything that they can get their hands on to launch themselves hundreds of years into the future in terms of technological advancement and development. Why develop things when you can just steal them from others, right?
But at least the communist dictatorship of China that has been using ideological subversion along with their cold war allies the Russians and Mexicans, to make Americans both hate themselves and throw open the gates to antithetical minded people, will surely be as benevolent and generous and self-sacrificing as the USA has been. They are communist after all, and we know that communism works really well due to all the examples of it working really well.
Of course they will. Americans have turned on each other in their period of greatest economic vulnerability, basically self-destructing while the Chinese government gladly hands out 5x your research grant from elsewhere to do that research in China. America's running on desirability to high-skill immigrants and existing ecosystems, both of which are fading fast. NIH is halving budgets while China is multiplying them by 10. The West is fucked, I just wish I spoke Chinese.
I up-voted you, even though we disagree a bit. A friend went to work in Hong Kong years ago because it was easier to get long term AI research money than in the USA. I think he is still happy with the support that he gets.
A little off topic: here in the USA, every medium and large company seems to be morphing into an 'AI company' which I think is generally a good thing since individually they can all take advantage of control over their own application stacks and data sources. Many people concentrate on the advantages of huge companies like Google/Microsoft/Facebook in AI development, but there is plenty of opportunity for smaller companies to enhance their own systems with machine learning.
I have worked in the field of AI since the early 1980s, and experienced the 'AI winter' which was a natural reaction to over-hyping of technologies like expert systems that required too much manual effort to write and maintain.
Deep learning applications are developed with less manual effort and produce much more value.
The 'AI world' is very different now than in the 1980s.
109 comments
[ 2.6 ms ] story [ 162 ms ] threadIf that's what gets you going: Russia could also match or beat America in AI.
Instead, what is needed is user labeled images, videos, text, speech and of course, simulations (games, VR) for virtual robotics and reinforcement learning agents. These kinds of data and simulations are open sourced and growing. Once there is a sufficient collection of training data, it's cheap to train your neural nets.
Even today, there is much more data than scientists care to use. For example, ImageNet is over 1TB, but a smaller portion of it is used routinely (it has 21841 topics but usually just 1000 of them are used in research papers). So it's not the lack of images that's keeping them. China would have no advantage here. Everyone has approximately the same level, the best paper of last year is the average of this year. That's about AI in general - but regarding self driving cars and ads Google is cooking in secret so it might be a little more ahead.
umm:
google image search, all html embedded/related metadata, adwords itself serving as a refined metadata/tagging and ontology database building engine, recaptcha 'how many buildings are in this picture' puzzles, android debug settings where things are uploaded, google drive cloud storage, gmail, the google speaker gizmo and voice activated phone agent, google voice voip, google chat, maps annotations to pictures/gps coordinates, etc etc etc and yadda yadda yadda...
not sure what else you need as far as those sources..
Are you saying no software has trade secrets?
Sure, you can go through some tutorials and learn to build a CNN, or even read more recent research papers and hit bleeding-edge ImageNet benchmarks, but nothing currently public is going to get you close to Google's SDC.
You can't pretend this is just about people either, India's at 1.3B and they're far from becoming a world leader.
common counterpoints: secrets can be operationally necessary, secrets can create economic value.
Not anyone can do AI at scale.
For that you'd have to wait until next year. Many low level skills necessary in ML are being automated. It's getting closer to being easily used by hackers.
A case in point. There is a conference, SYNC 2017, in Computer History Museum tomorrow (July 14th), organized by Pingwest, which itself seems like a Chinese Gov. front if you look at the team.
They have several good speakers, and some Chinese, Gov fronted, fake VC types.
Till 2 days ago, it was titled SYNC 2017 - Unbounded: Building Across Borders. Yesterday, it's title was changed to SYNC 2017 - US China AI Summit
It is pretty clear that if they had gone with the US China AI Summit in the beginning several of their speakers may not have agreed, but now it is too late for anyone to change mind.
Edited: To change from "Chinese" to "Chinese Gov." to clarify I'm referring to the Chinese Gov and not Chinese people
I'm not sure how those speakers couldn't just cancel, sure they'd lose any speaking fees but I've never heard of lock-in speaking contracts for conferences. Or if they care so much they could just modify their speeches to not supply anything of value. Or perhaps they should have done their homework ahead of time.
In any case I'd hardly call it a "nefarious trick" regardless of who was doing it. I'm hesitant to throw out "racist", but you seem to be leaning that way...
Nefarious or not, it does seem a bait and switch - won't you agree.
I had signed up to attend, initially, now I'm not so sure.
I have no love of the Chinese government, but this is a free exchange of ideas that is open to the public (for a $40 ticket). Given the actions of the Chinese government to date, if this is how they plan to get more technical data going into the future I'm hoping they do more of this and less hacking of proprietary/classified info. :P
I think most of these comments are referring to the former, a dislike/distrust for the PRC government, not Chinese people.
> reports in NYTimes and other reputable media
IIRC POTUS calls that fake news?
...you mean it's Chinese people? What on earth makes this a "Chinese front" rather than just a company with a Chinese team?
Of course not all of them are such, but many are. Just look at the team behind many of these, team's credentials and experience, and in many cases their addresses and actual offices.
It's important to note this because shutting down Silicon Valley/US access to Chinese VCs and companies won't necessarily just harm them. It would harm us as well. It does need to be looked at though, just carefully.
And by luck. If they didn't do it, someone else would have done it, whether it be US, UK, or Canada. Proof? Look at the developments and publications from that same team. There are not consistent breakthroughs.
This is similar to the stock market and investors. Someone will get lucky and win, but unless they can do it consistently, it's luck.
I do know a decent number of top Chinese VCs looking into investments in Silicon Valley. Those funds are typically started by Chinese entrepreneurs who made a fortune in things like the internet wave or real estates.
There do exist branches of Chinese government, or nation-owned organization making investments in US. But they are not the top VCs, and I haven't seen any of them taking a stake in a top US AI startup.
There are a couple of reasons why Chinese money is flooding into silicon valley. First, there is too much cash in the domestic market and not enough good tech investments to make. Second, there is a fear that Chinese economy might crash someday so having oversea investment diversifies the risk.
From what I know we don't have government background or funding. PingWest is a privately owned and funded media company in China. We run one of the most influential tech blogs there. Our US team organize SYNC events every year in the Valley and some pop up SYNC events elsewhere.
We changed the title simply because it fits the event and our speakers lineup better.
[1]: https://www.crunchbase.com/person/thomas-luo#/entity
[2]: https://www.linkedin.com/in/thomasluoyihang/
https://en.wikipedia.org/wiki/Yellow_Peril
'The Yellow Peril (also Yellow Terror and Yellow Spectre) is a racist color-metaphor that is integral to the xenophobic theory of colonialism: that the peoples of East Asia are a danger to the Western world. As a psycho-cultural perception of menace from the East, fear of the Yellow Peril was more racial than national, a fear derived, not from concern with a specific source of danger, from any one country or people, but from a vaguely ominous, existential fear of the vast, faceless, nameless horde of yellow people opposite the Western world. As a form of xenophobia, the Yellow Terror is the white race's fear of the rising tide of colored people from the Orient.[1]'
It's entirely possible that you have a valid point to make, and maybe it's even an interesting and important one, but maybe also take a second to consider if it might be motivated by racist tropes.
Why do you think this has anything to do with racism? Is anything claiming an Asian country is doing better than a white country Yellow Peril? If not, why is this an instance of it?
Indeed. Textbook projection
Rome burns and everyone is asleep or mad with partisan fever.
Helpful article: https://www.thebalance.com/what-is-the-debt-to-gdp-ratio-197...
It also doesn't sit in isolation. The BLS statistics add another 200 or so percent for about 300% debt-to-GDP when you include household, state and muni debt.
Unlike China, the US citizen lender argument isn't helpful. US debt holders are free to move their money into other assets.
The only argument in favor of the current debt levels is that no other nation can float enough liquidity to compete with US debt. For that reason alone the sky isn't falling.
The shortcomings for Social Security are easy to fix. I don't know about Medicare and it's actuarial prospects.
Crossing the 100% debt/gdp ratio is not a Rubicon. Lots of countries have done this and been none the worse for having done so. The U.S. famously exceeded 100% debt/gdp ratio during WWII. It was able to come down from that.
Current debt obligations for the U.S. are in low interest bonds. The borrowing of the money has not crippled the nation. Spending it poorly on stupid wars, a massive surveillance/security apparatus in the form of DOD, CIA, etc. has been a problem. Currently, as a percent of GDP taxes in the U.S. are quite low.
https://en.wikipedia.org/wiki/List_of_countries_by_tax_reven...
And what do people want to talk about? how bad (supposedly) US is. When US leads the world in innovation and protection against aggression from NK and Russia (and China), ruled by dictators that brutalizes its own people. Whataboutism.
https://fred.stlouisfed.org/series/GFDEBTN
Rubicon? I don't recall the sacred oath not to cross a 100% debt-to-GDP ratio. What's the importance of that particular round number, which many countries including the US have crossed in the past, except for rhetoric?
Finally, what makes you think corporations are holding cash waiting for fiscal stability, of all things? How would US dollars or bonds help them in case of US fiscal instability? Isn't it more parsimonious to think that they're waiting for return on investment? And that has much more to do with growth than with government debt.
10% goes to welfare. 6% goes to interest on debt. 16% goes to defense. 25% goes to CHIP/medicare/medicaid. (25% to social security, but that has a separate tax inflow and is a net positive to the budget.)
see: https://www.cbpp.org/research/federal-budget/policy-basics-w...
http://darribas.org/WooWii/Assignments/OriginalPaper.pdf
The broader point is, if I issue money and then issue obligations on that money, there's no practical difference between the two issues. That is, public debt of countries that are sovereign in their currency, and only issue debt in their currency, cannot have public debt overhang. It's just a red herring, based on a false analogy between governments (currency issuers) and households (currency users).
Michael Pettis regarding the research you cite states: "What is usually missing from such discussions, however, is an explanation of how too much debt leads to slower growth." ... "A country has too much debt when stakeholders begin behaving in ways that reduce growth. The controversy over the research by Rogoff and Reinhart should not be allowed to obscure the nature of the link between debt and economic performance."
https://www.project-syndicate.org/commentary/the-misleading-...
So, absent a proven behavioral line in the sand, we have only measures of debt to try to surmise at what point debt impacts growth and stability.
Its really chicken and egg.
Regarding specific claims in your linked article:
- Bond markets in such countries are subordinate to central banks; the most obvious case is Japan, where a greater-than-200% debt ratio is accompanied by, I believe, negative yields on public debt. Central banks (again, in countries that have nonconvertible sovereign currency) totally control short-term interest rates, and so indirectly control long-term rates. Creditors take the central bank's number and then risk-adjust and discount it for borrowers; for the government the risk is zero, so they just discount it.
- Ricardian equivalence is a stupid model. Have you ever made consumption or investment decisions based on a belief that public deficits will result in higher tax rates at some point in the future? I certainly haven't.
- "Workers organize to protect their jobs," but I guess the anti-union bent says a lot. Nevermind that spending equals income; a falling wage share is a far more plausible explanation for less-than-desirable growth in most situations.
- "savers [...] withdraw their bank deposits" The loanable funds model is not how banking works, but that's a much lengthier discussion. Suffice it to say, it's still the case that central banks set interest rates. I assume this is the reason the author brings this up, because otherwise the question is---they withdraw their savings and do what?
- Again, the only way the US government can default on US Treasury obligations is if Congress explicitly decides to do so. There's no mechanical or functional reason this might happen.
Historically, countries have not been sovereign in their currency. You'd have to be more specific about Argentina, though its biggest debt crisis was because its currency was pegged to the US dollar. Likewise, the current troubles in the Eurozone come from a lack of currency sovereignty. These things are unfortunate, and will probably lead to a dissolution of the EMU at some point, but they have no bearing on the US.
We can always print more debt instantaneously thanks to the FED. China can't print more humans.
We have our problems for sure, but china is facing far greater problems.
Given that in the next four years it is likely that more than half of the PhDs awarded by top US schools are awarded to foreign nationals[1], and our current government policy is to restrict immigration as much as possible, we are on a path to insuring that soon the world will have a much better pool of talent to draw on to do the research than the US will.
[1] https://www.washingtonpost.com/r/2010-2019/WashingtonPost/20...
Then there are cultural differences that could be at play concerning innovation. It is a general observation that east asian cultures tend to suppress risk-taking and innovation. Whereas, western cultures seem to promote risk-taking and innovation. That is not to say that there aren't risk takers in China and risk averse individuals in the West...just that overall there may be asymmetry. And, very little asymmetry can create very big differences.
So, have all the smartest people you want, doing all the research you want; but if they have no pressure to put that research to practice, nor a desire or support to take the risks associated with implementation then you might as well have nobody doing anything.
Just a thought.
The use of “misnomer” in that sentence is, itself, misnomer. What you describe is, perhaps, a mistake, but not in naming.
In the case of China, I don't think there's any question that it used to have higher barriers to innovation but I'm not comfortable concluding that this will be true for that much longer given how many people are trying to change it. Every time I read one of Bunnie Huang's gongkai posts[1] it really seems like the view in 20 years will be radically different.
1. https://www.bunniestudios.com/blog/?cat=20
That said, how long does it take for a culture to change part of its core structure? Cultures seem to me evolutionary and, aside from rare radical mutations, very slow to change. Then again, with over a 1.3 billion people there are sure to be plenty of sub-cultures with enough size to be a driving force for the whole.
It isn't that innovation necessarily comes from academia, but that academia is a reflection of how wealthy and educated the population is. Innovation comes from the wealthy and the educated. So as china gets wealthier, they will innovate more.
> It is a general observation that east asian cultures tend to suppress risk-taking and innovation.
That's an "observation" based on racism and nonsense.
> just that overall there may be asymmetry.
Based on what evidence? It's just silly things people say - "The west is more individualistic". A society based on "white supremacy" is "individualistic"? How is basing one's ideology on "race - a genetic/communal" concept individualistic.
All these things are essentially nonsense created by people who want to sell books/etc. There is no truth.
You could write a book saying the west is collectivistic and china is individualistic by cherrypicking any data.
> So, have all the smartest people you want, doing all the research you want; but if they have no pressure to put that research to practice, nor a desire or support to take the risks associated with implementation then you might as well have nobody doing anything.
What makes you think they won't?
Greed works on the chinese as much as westerners. The things you need to "innovate" are wealth and security. Innovation is a luxury. The chinese will innovate as they get wealth and security.
The chinese are risk averse and yet they love gambling. They are risk averse and chinese have immigrated all over the world.
This is correct, but that doesn't mean causation. Look at any civilization and look at what came first wealth or academia. It seems to me that wealth creates academia.
> That's an "observation" based on racism and nonsense.
Saying something is "racist" does not constitute an argument. I didn't say whether it was "good" or "bad" to have a culture that promotes or suppresses risk taking...no one can. It is not "racist" to say that two things are not exactly the same.
> The things you need to "innovate" are wealth and security. Innovation is a luxury.
Dead wrong. Wealth and security create complacency. "Necessity is the mother of invention" is not just some stupid quip.
>What makes you think they won't?
I don't. My entire post was a postulation, which is why I used phrases such as "it could be", "general observation" and "there may be", etc. It appears that you are merely looking for a fight instead of having a conversation.
Yes. That's was my point. You get wealthy, then you get "civilized".
> Saying something is "racist" does not constitute an argument.
No. But saying an "observation" based on "racism" is an argument.
> I didn't say whether it was "good" or "bad" to have a culture that promotes or suppresses risk taking...no one can.
Well you did imply it...
> Dead wrong. Wealth and security create complacency.
Not dead wrong. Dead truth. It can for some people, but it also gives people FREE TIME to pursue other things. I didn't say EVERYONE would become innovative. My point is that you need wealth and security to be innovative.
>"Necessity is the mother of invention" is not just some stupid quip.
Like all quotes. It is. And without wealth and security, not much time or resources to innovate.
> I don't. My entire post was a postulation, which is why I used phrases such as "it could be", "general observation" and "there may be", etc.
You don't but that's a lot of "implying" what you think.
> It appears that you are merely looking for a fight instead of having a conversation.
I would but you seem quite defensive. I wasn't calling you racist. I said the observation was based on "racism". Lumping 1.4 billion into one "stereotype" is the definition of racism.
Also, as I noted, I debunked most of your "it could be"s.
Competition is good.
Ironically, underlining in this article is that acknowledgement of such often unethical or act in doubt are perhaps essential and critical to AI's future development. To that, China is at a clear advantage than America, and the rest of civilized west world.
A strong economy is also necessary to propel the development, or Russia would also be at a advantage similar to China's, or North Karea's. But only China has the economy incentivization similar to West's.
Among all the things, this is the fundamental advantage China has over other nations. Chinese government can do things that no other are capable.
"Civilized"? Come on, the civilized "west" was performing human experiments. If china is doing it, then so are we.
Lets stop with the hubris. We aren't civilized. We are savages just like any other savages.
But at least the communist dictatorship of China that has been using ideological subversion along with their cold war allies the Russians and Mexicans, to make Americans both hate themselves and throw open the gates to antithetical minded people, will surely be as benevolent and generous and self-sacrificing as the USA has been. They are communist after all, and we know that communism works really well due to all the examples of it working really well.
A little off topic: here in the USA, every medium and large company seems to be morphing into an 'AI company' which I think is generally a good thing since individually they can all take advantage of control over their own application stacks and data sources. Many people concentrate on the advantages of huge companies like Google/Microsoft/Facebook in AI development, but there is plenty of opportunity for smaller companies to enhance their own systems with machine learning.
They have enough labor to do supervised learning on a grand scale that the U.S. can't compete with.
The U.S. can't compete.
I have worked in the field of AI since the early 1980s, and experienced the 'AI winter' which was a natural reaction to over-hyping of technologies like expert systems that required too much manual effort to write and maintain.
Deep learning applications are developed with less manual effort and produce much more value.
The 'AI world' is very different now than in the 1980s.