Many of the best AI researchers are Chinese. But they almost all work for US companies. Which is why we cannot mess with immigration.
This article is written for people outside the field? Baidu is alright but a bit of a shambles. All the did was build a dense GPU rack for RNN training. Their imagenet results were from cheating. The theoretically interesting work isn't really happening there.
The Microsoft paper was funny. It was PR BS that runs 10x real time.
Speech is nice, and iflytek does fine, but speech is becoming a commodity now.
The interesting work isn't happening in either of those companies and it's not happening on those problems.
This article was written by someone who didn't know the field interviewing someone who is trying to get his budget increased via fear mongering.
You mean US companies with international offices such as Geoff Hinton at Google Toronto and DeepMind owned by Alphabet in the UK? Not to mention Facebook AI Research in Paris. Yep. Canada and Europe sure are producing strong research for the US. Thanks for the talent!
Agreed! From my perspective (computer vision), I can think of quite a few Chinese individuals who are true leaders in AI at this point (often working for US companies).
In fact, given both Hinton and Bengio's being Canadian and LeCun being French, I would say US really in the game in terms of producing "AI grandmasters" (assuming that's their definition of grandmaster). Krizhevsky was studying in Canada. DeepMind is European.
From both the research and industrial perspectives, China's education system has well prepared students for the field of deep learning - the rigorous / analytic nature has proven a massive benefit when dealing with such large models and datasets.
>The new Chinese weapon typifies a strategy known as “remote warfare,” said John Arquilla, a military strategist at the Naval Post Graduate School in Monterey, Calif. The idea is to build large fleets of small ships that deploy missiles, to attack an enemy with larger ships, like aircraft carriers.
This reminds me of a famous story of an early AI system, which competed against humans in a table top war game focused around ship design.
>After weeks of experimentation, and some 10,000 two-to-thirty-minute battles, Eurisko came up with what would be the winning fleet. To the humans in the tournament, the program's solution to Traveller must have seemed bizarre. Most of the contestants squandered their trillion-credit budgets on fancy weaponry, designing agile fleets of about twenty lightly armored ships, each armed with one enormous gun and numerous beam weapons.
>Eurisko, however, had judged that defense was more important than offense, that many cheap, invulnerable ships would outlast fleets consisting of a few high-priced, sophisticated vessels. There were ninety-six ships in Eurisko's fleet, most of which were slow and clumsy because of their heavy armor. Rather than arming them with a few big, expensive guns, Eurisko chose to use many small weapons.
>In any single exchange of gunfire, Eurisko would lose more ships than it destroyed, but it had plenty to spare. The first battle in the tournament was typical. During four rounds of fire, the opponent sank fifty of Eurisko's ships, but it lost nineteen -all but one-of its own. With forty-six ships left over, Eurisko won.
>Even if an enemy managed to sink all Eurisko's sitting ducks, the program had a secret weapon -a tiny, unarmed extremely agile vessel that was, Lenat wrote, "literally unhittable by any reasonable enemy ship." The usefulness of such a ship was discovered during a simulated battle in which a lifeboat remained afloat round after round, even though the rest of the ships in the fleet had been destroyed. To counter opponents using the same strategy, Eurisko designed another ship equipped with sophisticated guidance computer and a giant accelerator weapon. Its only purpose was killing enemy lifeboats.
I've always wondered if a similar strategy might work in real life.
People in pretty much every country are doing some pretty amazing stuff. A lot of it you won't hear about if you limit yourself to press releases from venture capitalists.
>People in pretty much every country are doing some pretty amazing stuff. A lot of it you won't hear about if you limit yourself
I agree, and this is why I am unworried about the recent moves to cut government research funding in the US. There are plenty of other countries with great research institutions.
Look what happened when Bush pushed out most of the stem cell research -- research moved to Singapore and China. But the world still benefited, just not US companies.
Extremely impressive website with lots of fascinating examples. I assume the website was generated by the real-human-brain algorithm with the given task of "maximum impress site visitor". This will change everything!
I ran into it on some Reddit stream and thought it was funny as he claims AGI and he is everywhere asking/pleading for buyers. In the Netherlands we had Sloot[0] who claimed a similar impossible feat in another field. Well Sloot was actually claiming something 'more impossible'; I do think strong AI will happen but we are far off while Sloot claimed something impossible (Kolmogorov complexity is a good book about that). Sloot was believed by many and invested in. Also by people who should know better.
Normally I would say China, because it has a strong incentive to create an automated system that censors everything the leadership wants, and that manages "citizen scores", and spies on everyone at all times.
However, I wouldn't be surprised if the U.S. intelligence agencies want to do that just as much to Americans and the whole planet, but do it in secret.
"Many Silicon Valley firms remain hesitant to be seen as working too closely with the Pentagon out of fear of losing access to China’s market." This is an interesting assertion, but I'd like to see some evidence.
Not sure about this blanket statement, with Palantir being a shining example of why it's likely overtly wrong (DoD is most likely their largest source of revenue).
Byzantine hiring processes in the government itself doesn't help this problem. This isn't a politica statement, if you compare USAJobs to an average tech company you'll see an over emphasis on certifications and the like. There are reasons for it, but it doesn't help when decision makers or project managers don't know what real success looks like or how to even measure it.
I'm not sure that is a problem...historically the department of Energy (DOE) has lead Supercomputing efforts in the USA and they have focused on other domains of interest (particularly nuclear weapons research). This field is actually not unrelated, in terms of HPC capabilities, Numerical linear algebra, etc to the recent AI work. But physical simulation remains the top priority in part because AI is not seen as an energy technology.
I can state without hesitation that some of the greatest minds in Supercomputing and computational science are permanent members of the staff at these DOE research labs. These staff are selected much like the top AI researchers: academic pedigree and publications in top tier research journals.
Therefore, I think it is more a problem of application. Until government has a killer AI app, it will be hard to justify a massive investment in AI tech at scale. Contrast that with large companies who have already deployed AI tech in production...
That's because it's a classic refrain that most commonly refers to international competition more often than domestic competition, due to the obvious broad implications to national security and defense. It's not in any way limited to or specific to China - and it's even occasionally used domestically in reference to tech companies battling (see: frequent arms race references in regards to Amazon vs Google vs Microsoft in cloud).
The exact same language would be used if the US were competing against Russia on this, and previously was used when the USSR was the primary political competitor of the US. It has also been used to refer to competition with Japan, now and in the past.
NYTimes, 2005, vs the Japanese (and recently ascendant China):
"A global race is under way to reach the next milestone in supercomputer performance, many times the speed of today's most powerful machines. And beyond the customary rivalry in the field between the United States and Japan, there is a new entrant -- China -- eager to showcase its arrival as an economic powerhouse."
Do you have specific criticisms of the NYTimes piece? Reasons to think these particular claims are false? If you do, please do share, as it would be a substantial contribution to the conversation. Throwing aspersions or drive-by allusions of doubt is a primary reason why we're in the current situation where we can't seem to find agreement on so many things that should be facts.
Your links provide more support than you comments regarding folklore. Other cultures are no stranger to superstitions, quack remedies, and non-scientific religious practices.
China has also progressed quite handily economically and technologically, areas that don't admit to too much reliance on things that don't work. Cautious skepticism isn't the same as out-of-hand dismissal.
It's problematic to take scientific claims at face value, regardless of the claimants' ethnicity. It's also problematic to discount claims purely on the basis of the claimant's ethnicity.
AI/machine learning has slightly more accountability than many other scientific fields, in that unless you provide an implementation for people to inspect or your results are easy to replicate, people generally aren't going to be particularly interested for long. There are exceptions to that principle, like AlphaGo, but the tests of AlphaGo speak for themselves. It's a very results-driven field, where you're not going to be able to get away with specious claims for long.
This article is written for people outside the field? Baidu is alright but a bit of a shambles. All the did was build a dense GPU rack for RNN training. Their imagenet results were from cheating. The theoretically interesting work isn't really happening there.
The Microsoft paper was funny. It was PR BS that runs 10x real time.
Speech is nice, and iflytek does fine, but speech is becoming a commodity now.
The interesting work isn't happening in either of those companies and it's not happening on those problems.
This article was written by someone who didn't know the field interviewing someone who is trying to get his budget increased via fear mongering.