SVIEF Kai-Fu Lee Keynote: Era of AI, Rise of China, U.S. vs China, etc & All Star Panel Session (?)
Well respected technologist, entrepreneur, writer and AI researcher Kai-Fu Lee, PhD presented a powerful and very incisive keynote speech on September 29th at the SVIEF conference in Santa Clara, CA. The title of his talk was all encompassing and compelling: “Era of AI, the Rise of China, and the Future of Work.”
Author’s Note: I was so impressed with Kai-Fu’s talk, I’ve ordered his latest book, AI Superpowers, which is already on Amazon’s best selling book list.
Here are several important highlights of Dr. Lee’s SVIEF keynote:
1. Deep Learning (DL) is the biggest technological improvement in the 60+ year history of artificial intelligence (AI). DL is a network of highly connected neurons in thousands of layers that can, in a single domain, take a huge amount of data and train to recognize, predict and decide and synthesize at a much higher accuracy than humans.
2. DL is not human intelligence, it cannot think or cross domains. It has no strategic or creative thinking capabilities. But in a single domain, with a huge amount of data, it is beating humans in almost every task imaginable. For example, AlphaGo (a computer program that plays the board game Go) has beaten Go champions. In addition, we’ve seen DL used effectively for speech recognition (e.g. Amazon Alexa, Google voice search, Microsoft Cortana, etc.) and facial recognition. There are new beginnings of DL diagnosis of how to read MRIs and doing a better job of that than radiologists.
3. This amount of improvement is leading to what are four waves of artificial intelligence:
- Wave 1: Internet AI started in 1998. The Internet has more data than any other domain. With so much data, it enables Amazon to predict what you might want to buy. It powers Facebook to predict what you might want to read on-line. Similarly, all the American and Chinese companies (Alibaba, Baidu, Tencent, etc) , all the great AI companies of today are all Internet companies, because they have the most labeled data.
- Wave 2: Business AI started in 2004. Take banks, insurance companies, hospitals –they have amassed a lot of data in the past, they viewed data as a call center, as a legal requirement to archive. But now data has become a goldmine for them in various ways.
- Wave 3: Perception AI started in 2011- the ability to see and hear. Examples include: computer vision, speech recognition, speech synthesis, understanding all combined together. It also can be viewed as digitizing the physical world.
- Wave 4: Autonomous AI (self driving cars, autonomous robots, etc) started in 2015. In this wave 4, AI becomes autonomous in its ability to move around and manipulate sort of like having hands and feet. That will usher in an era of autonomous vehicles and robotics. Autonomous vehicles will bring about a huge transformation, especially the dis-incentive to own a car. With safer autonomous vehicles, the natural next step is humans won’t be allowed to drive anymore.
4. To make AI work, we need the following things: a lot of data that is tagged within a single domain, a lot of compute power, and some AI experts to work on it.
AI is not perfect — you can’t make it do perfectly unsupervised learning. You can’t make it learn on very little data. You can’t do AI with very little compute power.
But once you have those in place, AI can be effectively applied.
5. U.S. Leads China in Top Researchers, Patents, and AI Talent (and will likely continue to lead in AI research in the near future).
6. Chinese Miracle of Last 10 Years with fast product/service iteration, intense competition, user acquisition, accelerated growth, high return on investment in a huge market.
7. In 2018: U.S. and China Have Become Parallel Universes:
US Model: Breakthrough Technologies, Vision-driven, Light, Globalized.
China Model: Fusion + speed, Applications, Result-driven, Heavy, Localized.
8. Investment in China: A lot of money and capital investment went into China with smart VCs helping smart entrepreneurs build products and companies. And those products actually are so attractive they brought more Chinese users on the internet. And this loop kept going and going for the last 10 years taking China from 150 million users to about 800 million users by far the largest user base in the world. And this loop has created something that we never thought possible — a system that parallels the Silicon Valley.
9. The only way to succeed in China is to find a business model that is impregnable. In other words, build a business that’s uncomfortable. Chinese companies kept improving going from copying from the U.S. to inspired by us and then leapfrogging the U.S. For example, WeChat (messaging app) is better than WhatsApp and way better than Twitter. But even more exciting is the third ladder where Chinese companies are brand new innovation, this Chinese model of building impregnable businesses have reached new heights, so that these brand new companies are being built.
10. China Advantages over U.S. in AI:
Advantage 1: Chinese Product Innovation has Caught up with U.S. Pure Chinese Innovations Have Arrived
Advantage 2: Tough Market Begets Tough Entrepreneurs
Advantage 3: China’s AI Capital Leads the World. 48% of global AI investments were made in China; 38% in U.S., 13% other countries. SOURCE: CB Insights 2017 Global Artificial Intelligence Investment
Advantage 4: AI Moves into Era of Implementation
Advantage 5: China is the world leader in amount of Data (like Saudi Arabia is the country with the most oil for export). Massive Data is Critical for AI Product Success– even more important than algorithms. AI algorithms are generally shared, and it is up to the speed, execution, and size of the data that determines how companies will benefit from AI implementations.
11. U.S. Advantage over China in AI: Early Adopters, Expert is King (vs China which is Application Driven and Data is King)
12. Who’s ahead in AI, mobile and Internet: Dr. Lee thinks that generally U.S. is a little bit ahead today. But China will probably be ahead in four or five years. This is not about research. This is about implementation.
U.S. will continue to be ahead in research for the next 10 or 15 years, because that lead is very difficult to overcome.
But this is not a zero sum game. U.S. VCs fund U.S. companies that develop products for us customers, whereas Chinese VCs fund Chinese companies who develop products for Chinese customers (domestic market). The two countries are not going after the same market.
“When a Chinese company wins, a U.S. company does not lose. When a U.S. company wins, a Chinese company does not lose. So I think the sentiment behind the current some of the current rhetoric is not correct. This is truly not a zero sum game. This is merely a keeping score of how far ahead each technology might get. So with China and us both pushing forward AI, I think AI will make a lot more progress than internet and mobile because those only had one engine the U.S. pushing forward. And there are a bunch of other reasons such as the seven cloud giants (Amazon, Google, Facebook, Microsoft, Alibaba, Baidu, Tenent) hiring people, and training people with large amounts of data VCs being devoted to AI.”
- Embrace AI – it saves us from repetitive work and pushes us to do what human is called for.
- AI cannot create ideas or thoughts. We are the masters and should be responsible of how to use AI.
- (via Twitter) How U.S. can stay ahead in AI: 1) double AI funding, 2) increase AI professors pay, 3)offer green cards to all AI PhD’s.
“So going forward, I think AI is electricity in the next 20 years, there will be huge opportunities and challenges. But I want to take us a moment into (the next) 50 years. When we look back ignore for the moment all these job displacement opportunities, I like to leave you with two thoughts. The first thought is that AI is serendipity. It is here to take away the routine jobs so we can really spend time on what we love and what human beings are on this earth for. And secondly, for those worry about AI causing problems. Just keep in mind AI is just a tool. It possesses no creativity. We (humans) are the Masters. We are the ones that have free will. And it’s going to be up to us humans to write the ending to the story of artificial intelligence. Thank you.”
In a post SVIEF conference email exchange related to AI’s use in telecommunications applications, Kai-Fu wrote this author:
“Thanks — this is not my major area of expertise. But clearly communications in autonomous vehicles, IoT, and 5G, when combined with AI, will be a great combination.”
“On China catching up, it will be in technology related to the Internet, Mobile, and AI.”
SVIEF All Star Panel:
Kai-Fu Lee: “I feel a sense of social responsibility to tell people that as AI advances, job displacement is a serious issue. And I think I thought very hard about various solutions. I looked at universal basic income, I don’t think that’s going to work. I don’t really know what will work. But I do think generally, it’s in the direction of creating more empathetic jobs, because there should be a large enough pool of them, if only we would care about them, pay more for them. And that can hopefully lead us to a good ending.”
SVIEF All-star panel with VC Tim Draper, AI Rock Star Kai-Fu Lee, and Stanford Physics Prof. Shoucheng Zhang
Due to time and space constraints the above panel session may be summarized in a follow on article, provided there is sufficient reader interest.
Please email me at: [email protected] if you’d like me to write such an article.
Transcript of Kai Fu Lee’s keynote (via speech recognition):
5 thoughts on “SVIEF Kai-Fu Lee Keynote: Era of AI, Rise of China, U.S. vs China, etc & All Star Panel Session (?)”
Huawei lays out AI roadmap: Huawei unveiled its AI strategy and AI portfolio at the recent Huawei Connect conference in Shanghai.
“AI offers new business opportunities, strengthens our products and services to become more competitive and future-oriented, and improves our internal operational management and efficiency,” said Eric Xu, rotating chairman at Huawei.
10 changes shaping the future:
Huawei has identified 10 changes that will help shape the future of AI. They are faster model training; abundant and affordable computing power; AI deployment in different scenarios and the protection of user privacy; data-efficient, energy-efficient, and explainable algorithms; AI automation especially during processes like data labeling, data collection, and model design; practical applications to meet the needs of real-world execution; real-time and closed-loop system; multi-tech synergy between AI and other technologies; one-stop platform; and talent availability.
Five areas of AI focus
To pave the way for these changes, Huawei has unveiled its AI strategy in five areas of focus—AI research investment, full-stack AI portfolio, open ecosystem and talent, existing portfolio enhancement, and operational efficiency in Huawei.
“Our AI strategy is to invest in basic research and talent development, build a full-stack and all-scenario AI portfolio, and foster an open global ecosystem,” said Xu.
Huawei’s full-stack AI portfolio includes chips, chip enablement, a training and inference framework, and application enablement. Its portfolio can be deployed in various scenarios including public clouds, private clouds, edge computing in all forms, industrial IoT devices, and consumer devices, according to Huawei.
At the chip level, Huawei announced the launch of two chips—Ascend 910 and Ascend 310, which are touted to accelerate AI adoption in different industries. Xu touted that Ascend 910 has the world’s greatest computing density in a single chip, which can be applied in large-scale distributed training system. Ascend 310 is an efficient AI chip for low-power computing. The chips will be deployed to offer cloud services to customers next year. Its AI portfolio also includes an AI acceleration card, AI server, and AI appliance.
Huawei’s AI Program & Platform:
Huawei has released its AI Developer Enablement Program at HUAWEI CONNECT 2018. The program will help Huawei collaborate with developers, partners, universities, and research institutions. Huawei will use the program to build a better development ecosystem that can support AI resources, platforms, courses, and joint solutions. Huawei aims to work with partners to build an affordable, effective, reliable, and inclusive AI ecosystem.
According to Zheng Yelai, vice president of Huawei and president of Huawei Cloud BU, Huawei’s AI Developer Enablement Program offers a platform for technical communication, talent training, and innovation to developers, tutors, and Huawei partners.
For Developers, the program covers the following:
-20-hour free introductory training
-Three-week beginner AI training camp
-AI developer contests
-Innovation incubation camp for top talent to help transform R&D success into commercial applications
Huawei will support its partners by:
-Building joint solutions based on Huawei’s AI computing platform and development platform to promote AI application in multiple industries
-Setting up an AI promotion alliance to build a joint innovation lab
-Providing 1000 sets of free development environments, including development modules and boards
-Providing the first 20+ partners with expert resources, supported joint solutions, and extra support for AI product launches and technical implementation
-Sharing market resources with partners, and cooperating closely to further expand the market
-For universities and research institutes, Huawei’s AI talent development plan includes:
-Investing over 140 million USD (including HUAWEI CLOUD AI resources and AI suites) in AI talent education.
-Cooperating with universities and scientific research institutes to develop AI courses, publish text books, and support scientific research and talent training.
-Helping universities and research institutes build AI colleges and institutes, assisting in building AI labs, and participating in AI related programs held by the Ministry of Education.
-Helping universities train AI teachers. Students can take exams to obtain Huawei AI certificates.
-Supporting the participation of universities into Huawei cloud open community, and building a platform for communications between universities, research institutes and Huawei AI experts.
-Offering universities and research institutions the computing power and technology on Huawei’s AI platform, to promote scientific research and exploration in the AI field.
At present, Huawei has started developing AI talent at eight universities in China, including Institute for Interdisciplinary Information Sciences of Tsinghua University, the University of Science and Technology of China, Zhejiang University, Shanghai Jiao Tong University, Nanjing University, Southeast University, Xidian University, and the Institute of Acoustics of the Chinese Academy of Sciences.
Thanks for the great article Allan!
WSJ: Huawei Steps up AI game Challenging U.S.
by By Dan Strumpf [email protected]
Huawei Technologies Co. unveiled two new computing chips aimed at powering artificial-intelligence applications, marking the Chinese telecommunications giant’s first major push into high-end technology dominated by U.S. chip giants.
Huawei’s Ascend line of semiconductors includes a chip that is installed on servers and performs complex AI tasks like programming algorithms, as well as a second chip for more routine functions on smartphones and other devices.
With the AI chips, Huawei, the world’s biggest maker of telecommunications equipment and a major smartphone vendor, is challenging American companies like Nvidia Corp. , Intel Inc. and Qualcomm Inc.
The new components align with broader efforts by China to reduce its dependence on advanced U.S. technologies and develop such products domestically. Under Beijing’s Made in China 2025 development plan, semiconductors and AI have emerged as key areas that authorities want to develop at home.
“Computing power is the foundation of AI,” said Eric Xu, Huawei’s chairman, at a conference in Shanghai on Wednesday. “We need to provide more abundant and affordable computing power.”
Other Chinese companies are also pouring resources into developing AI chips, which typically feature unique designs optimized for processing large amounts of data at once. Last month, e-commerce giant Alibaba Holdings Ltd. announced plans to launch an AI chip next year, while Chinese startups like Bitmain Technologies Ltd and Cambricon Technologies Corp. are also working on such components.
Huawei already manufactures an AI-enabled chip for its smartphones under its Kirin brand. The company is likely to find a willing market for the new components in its home market of China, but it will face challenges in unseating rivals elsewhere around the world, said Mo Jia, an analyst at technology consulting firm Canalys.
“Huawei is likely to use a lower-price entrance strategy similar to its infrastructure or server business” to break into the market, Mr. Jia said. “In the future, I think Huawei’s AI market will mainly be in the Chinese market.”
Currently, the market for powerful chips capable of advanced AI processes like deep learning is led by American companies like Santa Clara, Calif.-based Nvidia. Google also offers a similar chip called a tensor processing unit, while companies like Qualcomm make AI chips built into devices.
However, unlike those chip makers, Huawei won’t be selling the chips directly to customers, Mr. Xu said. Instead, the chips will be sold as a part of servers, modules and via its existing cloud-computing business to Huawei customers, he said.
The AI effort follows success for Huawei in its mobile-phone business, which earlier this year overtook Apple Inc. as the world’s No. 2 vendor of smartphones after Samsung Technologies Co.
At the same time, Huawei is weathering increased scrutiny this year from authorities in the U.S., where its telecom gear has been banned due to security concerns. Such fears led Australia to bar Huawei from its fifth-generation wireless rollout, while officials in Japan are studying similar moves. Huawei has long said it isn’t a security threat and that it is owned by its employees and operates independently of Beijing.
Comments are closed.