Kai-fu Lee: Ten years later, China AI market will occupy half, I have returned 3 o'clock block chain Group _ Lee

Author | Obviously

March 7, the Great Wall in Beijing held the tenth session of "Gmic Beijing 2018 conference and AI theme Salon" activities. It is reported that this year's gmic Beijing 2018 will be held in April 26-28th in the Beijing National Convention Center, with "AI" to give birth to all things as the theme.

Sharon, the founder of the Great Wall, Li Kaifu and the chairman of Innovation works, Lee Kai-fu, Horizon founder and CEO Yukei, Shengjing Network co-founder and chairman Peng Zhiqiang each issued a view of AI:

Lee: The next ten years, China's AI will account for half of the world;

Yukei: Within five years, the pattern of the field of image recognition will be finally established;

Peng Zhiqiang: There is no doubt that China's AI market will surpass that of the US in ten years.

The chef joked about the theme of this Congress: "I have been advised that this year's conference theme changed to block chain, not AI life." "To this, participate in salon activities of Kai-Fu Lee Voice said:" I have returned 3 o'clock block chain group.

▌ai can be divided into four waves

On the salon, Kai-fu Lee from the Innovation Workshop Investment Analysis Point of view, analysis of the AI level of four waves, and think that in these four waves China and the United States will prevail.

First wave: Internet AI

The biggest numbers are internet companies, where China has more data than the US, and the US faces different regulations on data, particularly the EU's privacy regulations, which are a challenge for the United States. What we know best about Chinese Internet companies that use data to create value is that today's headlines and quick-talk are not going to be as good without AI these years.

So I think for internet companies, the past must be the United States leading, but China is fast overtaking, five years later, China's AI in the Internet to create the value of the largest. Of course most of these may not be the natural language you're talking about, no one is driving that sci-fi thing, it's turning a lot of Internet data into value.

Second wave: Commercial AI

Business AI is the bank, insurance companies, retail, also including hospital cancer, etc., a large number of stored data with AI activation, in its business process to create value, I think China overtook the United States is a certain degree of difficulty. Because the United States in the past more than 10 years, more than 20 years of its data warehouse and so the technology is very mature, its data is immediately available. China has also built several big companies, their data not to be said to be structured, and even have problems cleaning.

Of course, we think that there are still great opportunities in traditional enterprises, especially in the retail industry. We think of the unmanned store will first think of Amazon go, in fact this is not the first product, the unmanned store one-step, there is a person without this is not really our goal, we want to see how to make the customer experience better. So I think that China's attempts at a variety of hundred-way unmanned stores will allow China to overtake the United States in this respect. But as a whole, I think the United States will stay ahead of the AI that launches the business data.

Third wave: Perceptual AI

I think in the whole cancer medical data or in face recognition, object recognition, China will capture more data. For example, putting a lot of cameras in a classroom or retail store can help increase the user experience, in this way, the exchange of personal privacy for a higher user experience, the Chinese are relatively acceptable, the United States is relatively less acceptable, the Europeans are very disgusted, so this will lead to China in this respect can be driven faster.

There are certainly more data on cancer in China, but China's data are not being collated. There are also a lot of Chinese medical data that are not as accurate as the top American cancer hospitals are labeled. When the data can not achieve a certain degree of precision, the more the better, but also need time verification. Overall I'm as optimistic as Lurkey, so in the third direction, voice and video, perception I think China will certainly overtake the United States in five years.

Fourth wave: Automatic AI

At this level I think China will also have a considerable opportunity. Just as Yukei said, we are still a long way from Google Waymo, at least two years away. Can it roll faster because of more data, more capital, more entrepreneurs and more government support? There are also many policy issues in the United States and Europe, such as a tangle of ethical issues.

How do you decide, on autopilot, why you can make a left-turn choice and bump into a baby without making any other choices? I think that the Asian government will feel that the overall loss of our lives has become less, and that the health damage has become lower, so we have to promote. Western thinking may be more because every life to explain to me very clearly, even if you save 10 people, but affected three people, these three people explain not clear, that 10 people I do not care. This kind of policy and the human environment and social habits, can make AI push some obstruction.

So my conclusion is that in the fourth aspect, in robots and unmanned driving, although China has a considerable distance from the United States this year, I think five years can be leveled or even not ruled out.

▌ ten years later, Chinese AI accounts for half of the world

Then the question for the chef: Mobile Internet 10 years we are half, ten years after the artificial intelligence era China is not able to achieve what level, can reach half. The three-bit answers were given below.

Kai-Fu Lee: I think from the application and market capitalisation should be more than half, maybe 60% or so. But on the processor side, in the operating system, the top scientists may not be able to exceed. This is a bit like mobile internet, but better than before, because the mobile internet era, "top scientists" China may not have 1%. Now we have 5% to go up, so there will always be some progress.

Yukei: The development of artificial intelligence, in the words of Buffett is very long snow road. I think the next 10 years of development is only the beginning, perhaps 20 years before the dust settles. Because so far, whether in the image recognition, or speech recognition research and progress, is a very elementary artificial intelligence part, still far from reaching decision and cognition. But in the field of image recognition the war will end within five years and the pattern will be established.

Peng Zhiqiang: In the AI field we see the valuations of Chinese companies rising fast. Companies in Israel and the United States are much slower to value, including a market-driven move that is less ferocious than China's. So I think, from the market point of view, in 10 years after the Chinese market should say that there is no suspense in the AI field is more than the United States.

Of course, this is not just AI, probably in 2030 China's economic total should be the same as the United States. So the permeability of China's new economy is still very objective. Plus we are more aggressive in the "all the way". So I think in general China is going to overtake the United States in these new areas of AI.

China is characterized by the fierce market, the government's power is also strong, aggressive, so that these two forces together, in the emerging economy we do have a clear advantage. But it is true that we have a long way to go in terms of fundamentals, from scientists to the underlying technologies of the whole foundation, just as the teacher told us.

▌ future readings will mostly be written by AI

Salon scene, watching live friends asked questions: Ai How to change the media industry. Kai-Fu Lee replied: "The next 5-10 years, we see a large department of content will be machine-written."

Kai-Fu Lee: The media is basically the content of the production and selection of content. There was a big controversy in the United States over the selection and sequencing of content on Facebook, and whether AI replaced traditional media. What we've seen is that Zuckerberg made a concession because he insists that AI editors can do very well because it's the result of a user's reading click. The American media have assembled to attack Facebook, saying that you look at what you have lined up and which one has meaning.

This needs to have a certain balance media spirit, the first user need to get what knowledge, but also take into account the user clicks. I think it's a pity that Facebook has so many AI Daniel not making good algorithms. Because this is something that can be done. The future also needs to be constantly groping to solve, because pure by clicks, make out may be filled with the title party or bad content. The ones that should really be read now seem to need to be edited manually. I think there should be a decision-making mechanism to combine the two.

Future short content will be machine-generated. If the future is written in template-type content, such as yesterday's sporting events, your work will certainly be replaced. So I think that journalists and content workers must be to write a more in-depth, more perspective, more attitude, is the robot simple single domain thinking or patchwork can not write out. But even so, I think that in the next 5-10 years, most of the things we read are machine-written, the best things to read or manual.