What is Kai-Fu Lee’s judgment and expectation on the development of AI?

The DeeCamp 2019 AI training camp hosted by Sinovation Ventures has officially started and is accepting online registrations from now on. DeeCamp is committed to cultivating AI application talents who are good at using AI technology to solve real-world problems and meet the needs of the industry. It has been held for the third time and is one of the most mature AI training camps in China.

Kai-Fu Lee, Chairman and CEO of Sinovation Ventures, will also serve as one of the mentors for this year’s DeeCamp. At the launch ceremony, Kai-Fu Lee shared some of his views on the development of AI:

1. AI will usher in a fully automated and intelligent era, and will directly bring in RMB 100 trillion in GDP by 2030.

2. AI is extending from AI B2B to +AI in all industries

3. Traditional companies need a CAIO to plan how to use AI

The following is the full text of Dr. Kai-Fu Lee’s speech:

What I want to share today is that AI is now facing a very critical node. DeeCamp started this project two years ago, and we can see how forward-looking it was at that time. Moreover, it will play a very important role at this node today.

How much is AI worth? PwC conducted a study last year and told us that AI will bring a net increase of about RMB 100 trillion to the global GDP after 2030. Last year, the GDP of China and India combined was only this number, so AI is a huge increase. Later, McKinsey and other institutions also saw such a big opportunity.

So can such a huge value be created in 11 years? We can think about the fact that the entire AI strategy of Sinovation Ventures is based on these four waves. The first is Internet intelligence, which has already begun and is deepening. We have also seen that the world's greatest AI unicorns are basically Internet companies, because the Internet has the most data. Business intelligence has also begun, and we have seen many commercial AI unicorns, including in China and the United States.

Next is the intelligence of the physical world, which means capturing the vision, hearing, voice, and video that the eyes and ears can see, and then using them to make smarter facial recognition, voice recognition, unmanned conversations, and unmanned store applications.

The fourth wave is fully automatic intelligence. When artificial intelligence can act, pick up things, do things on the production line, and have smart agriculture, unmanned driving, and robots, these four waves are happening at the same time. Below I have also listed the fields that will bring great opportunities and disruptions due to the technology and data brought by these waves. Most of what we see now are Internet applications, and we see many hopes to disrupt the industry. We also see that today, because of artificial intelligence, our lives have become more convenient. In the future, we can see that if it is said that 100 trillion in value is generated, it is not difficult to imagine that if these fields are empowered by AI, added value, or disrupted, it will definitely bring at least 100 trillion in value.

What kind of transformation will AI go through to bring about this day? We can probably see four cycles.

The first batch was AI companies around 2012. The day we invested in Megvii marked the beginning of the era of AI investment and entrepreneurship in China, except for iFlytek, which had already gone public at that time. So this wave of companies should have started in 2012. This batch of AI companies, including Megvii, SenseTime, and Yitu, are all technology-driven, with PhDs as CEOs and technology-oriented, slowly looking for applications. They have found applications and created great value, but the value they have created is still very different from the 100 trillion mentioned earlier, or a difference of four or five orders of magnitude. How will AI go in the future to bring 100 trillion in value? It is definitely not to create 100 or 1,000 Megvii or Yitu, but to take a newer route. We have just seen the various fields empowered by AI. These fields are actually related to scenarios, implementation, and business. Therefore, AI itself is a B2B business. Only by empowering existing businesses with AI can we create greater value. It is not just about making very good online apps, nor is it just about letting Internet giants make more money, nor is it about becoming a pure AI technology company.

The mainstream we see now is AI B2B, which is why after the establishment of Sinovation Ventures, the first company to be spun off, AInnovation, is the representative of AI B2B. It creates greater value. In the first stage, you work on technology, you make products, you understand the scenarios, and you find one customer at a time. The second time, you want to work on an industry, maybe you are working on the retail industry or the manufacturing industry. You can first get a very good retail customer, such as AInnovation got Yonghui, or a very good manufacturing user, such as AInnovation got Foxconn, and Mars, who is present here. After helping these three great companies to make certain solutions, you need to expand it into products that can radiate more retail, manufacturing, consumption, and industries. At this time, the value you create is not just for one or two companies, but can cover the entire industry. This is the second stage.

Next is AI+. Do we still remember Internet+? We have forgotten it. Why have we forgotten it? Because the Internet has been added to all industries, and Internet+ has been successful. So what is the concept of AI+? It is the same as Internet+. Traditional companies also need to consider how to do AI. I mentioned in an interview in the United States that many traditional companies need a CAIO. Your company really needs to have an AI talent to help you plan where your AI should be used. You even need an AI department to help your entire company from operations to planning, to development to products, to manufacturing, to logistics, etc., to see where AI can help you empower the greatest value.

I believe that in the near future, a group of entrepreneurs will come out to debate whether it is AI+ or +AI. It doesn’t really matter, but I personally think it is mainly +AI. Traditional companies must have the motivation to do so. First of all, traditional companies that do AI+ will not only survive, but may also become the leader in a very fragmented field, because their profits will be improved, their market share will increase, and in the end, others may not survive, which will bring about a very big disruption and efficiency improvement. The 100 trillion just mentioned, when it comes to B to B, it will be raised to another level.

Going further, we come to the state of the Internet today. Every company has AI. In this era, AI is everywhere. Every department that needs AI must have AI, and data can be collected when data needs to be collected. Moreover, the application of AI has become very easy.

Let's go back to the first stage. When those mysterious Internet companies came out, everyone thought they were mysterious and it was very difficult to understand the Internet. Gradually, the Internet became accessible to everyone, with websites and search engines. Slowly, traditional companies began to use the Internet, and finally the Internet became ubiquitous. AI will follow the same path. AI and the Internet are two equally great technologies. So what needs to be done in this process? When only a few people understand the Internet, there is no way to popularize the Internet, and even no way to do Internet+.

Similarly, AI engineers may be the most scarce animal in the world in 2012, with only 1,000 of them, so every one of them can become a CEO, which is very scarce and very important. But in the B2B era, there will be tens of thousands of them, hundreds of thousands today, and even millions by 2020, and almost every project can do AI. What we see is that AI will gradually enter the mainstream from a so-called black technology. Of course, AI is definitely not mainstream today. If you find an engineer and ask him to do AI for you, he will definitely not be able to do it. It is still a very difficult technology, but it is no longer a black technology in the hands of 1,000 people. This process will make AI continue to become mainstream and create more value. This means that whether it is the government, the industry, or the investment company, making AI mainstream and enabling a large number of engineers to master AI technology is an important factor in the development of a country or an enterprise.

There are various studies on the Internet, which tell us that the demand for AI is growing very fast, but the supply is insufficient to meet the demand, and companies are also competing for talent with high salaries. We have contacted many companies, and they basically told us that they still can't recruit enough AI engineers. Who needs AI? AI graduates today have a lot of choices, including foreign companies, unicorns, startups, etc. Every company is a great company. And not just three, but 20 to 30, so everyone needs AI engineers. How do we think about the demand for AI from the perspective of the country? When AI reaches the next stage, which is the era of AI+, the needs of these companies cannot be met. What should traditional companies do? If traditional companies want to find an AI engineer, where can they hire people? The demand for talent is very parallel from the top of the pyramid to the bottom. So Sinovation Ventures saw this demand two years ago, so we created the DeeCamp program. We are more picky than Harvard. We only accepted more than 300 out of 8,000 applicants. This year, we accepted more than 600, and it is estimated that there will be 10,000 to 20,000 applications.

We have a very unique training method. We train a group of AI engineers in five weeks through academic + industry + hands-on training. Some of you may ask how five weeks is enough? We have a magic formula. If you are a top computer science student who knows a little about AI, five weeks is enough for us.

It was definitely not enough in 2012, because at that time AI open technology, platforms and tools were not enough, but it can be done today. And I recently talked with one of the world's largest fund managers about this training. He was very envious and asked why my country couldn't have such an institution. I said sorry, your country only produces 1/6 of the engineers in China. And our Innovation Works, plus the Chinese government's recognition of AI, as well as the brands of Innovation Works and partners, can really recruit China's top engineers and computer science students. This is the success of DeeCamp.

It is not just because the curriculum is well designed, it is not just because we were born at the right time, it is not just because AI just joins the industry demand of B to B and AI+ era, because China has a large number of engineers and engineering students. Chinese universities do not have very deep courses in AI, but they are already world-class in computer science. Under the leadership of top role models in various industries and with the support of the government, Chinese students all want to go the AI ​​path. It is the right time, right place and right people that give us this opportunity to select 600 lucky people. This is a public welfare training and we do not charge any fees.

You can see several things from what we are doing. First, from the growth of 30, 300, and 600 people, we can expect how big the number will be before 2030. Second, we also hope that our courses will be adopted by more institutions and universities, so that there will not only be 600 new forces in China, but even 6,000 or more. Third, we don’t just want to train these hundreds of thousands of engineers. We believe that the top ones picked from the top will be CTOs, CEOs, or senior engineers in the future, and they will generate huge value. Finally, I believe that China will quickly maintain its world-leading status in AI development because of the large number of engineering graduates, coupled with top-level training.