Robin Li Visual China Infographic
"Before 2027, China's first-tier cities no longer need to restrict purchases and driving, and by 2032, the congestion problem can basically be solved by improving traffic efficiency."
On January 10, Robin Li, founder, chairman and CEO of Baidu (NASDAQ: BIDU, HKEX: 9888), made the prediction at the Baidu Create AI developer conference. He judged that the hallmark of the fourth technological revolution is deep learning algorithms, and said that software stuck in the neck is just as important.
Li Yanhong said that he sees intelligent transportation as an "intelligent dispatching system" that can improve traffic efficiency by 15% to 30% through intelligent transformation of the transportation network. Baidu's intelligent transportation solution has landed in 63 cities across the country, and the Ministry of Transport has officially listed Baidu as a pilot unit for a transportation power.
It is worth noting that at the Baidu World Congress in July 2022, Robin Li also made a prediction on this topic. At that time, Li Yanhong said that within 4 years, first-tier cities will no longer need to restrict purchase and driving, and intelligent transportation is expected to reduce traffic accidents by 90%.
At the Baidu Create AI developer conference, Robin Li's speech revolved around "growth", in the past year, people have discussed less about "wind outlets" and paid more attention to the health of growth, and around "growth", he judged the signs of the fourth scientific and technological revolution.
Li Yanhong judged that the hallmark of the fourth scientific and technological revolution is the deep learning algorithm, which can bring efficiency improvement and drive greater economic growth than many people think. Major innovations related to deep learning, including intelligent dispatching systems in the fields of autonomous driving and hydropower, will have a major social impact like inventions such as automobiles and the Internet, and are major innovations. "So, it's technological innovation that drives big growth." Li Yanhong said.
So where does the innovation itself come from? Robin Li summed it up as "feedback-driven innovation".
"Innovation is not behind closed doors, innovation is the opportunity to enter the market, constantly obtain feedback from users and customers, and feel the 'feedback' to cross the river." Li Yanhong mentioned that many times, the actual development path of things is very different from the original idea. The development of technology, there is no navigation map, only a compass. In the case of roughly correct direction, based on practical feedback, step by step iteration, in order to run valuable innovations.
However, technology strategy means a long cycle, and a long cycle cannot be smooth sailing.
Li Yanhong also talked about his return to China in January 2000 to start a business. Two months after its founding, the Internet bubble burst, the world lost 8 trillion market value, and many people were shaken by the difficulties and chose to give up and choose to leave. Finally, those who persisted in surviving the cold winter achieved the golden decade of Internet development. "As artificial intelligence moves from today to the future, such ups and downs will still occur. On the positive side, in the past year, artificial intelligence has changed directionally, both at the technical level and at the commercial application level. The most representative of the changes at the commercial application level is autonomous driving. ”
Li Yanhong again mentioned at the meeting that after L2, the first to enter commercial use is likely to be L4, not L3, because the definition of accident liability for L2 and L4 is clear, and the definition of accident liability for L3 is unclear, so it will take longer to popularize.
Li Yanhong has said in public that autonomous driving with scientific thinking to do, is to "step to the sky", for example, to achieve L5 and then expand the scale for commercialization, in a few decades may not be able to achieve. If you use engineering thinking to do it, it is to gradually realize automatic driving in some scenarios. For example, first achieve L2, realize the ability to follow the car on the highway, change lanes, and automatically park in the parking lot, etc., and then try to gradually move from L2 to L3, L4, and L5.
According to the information disclosed by Baidu, at present, Baidu's L4 autonomous driving has landed quickly, as of September last year, the cumulative order has exceeded 1.4 million, operating in more than 10 cities such as Beijing and Shanghai, and opening fully unmanned commercial operation in Chongqing and Wuhan.
In addition, Li Yanhong mentioned that what Baidu does can be divided into four layers: chip layer, frame layer, model layer and application layer. From the high-end chip Kunlun, to the flying paddle deep learning framework, to the Wen Xin pre-training large model, there are key self-developed technologies at all levels, and there is a lot of feedback between each layer, and end-to-end optimization is achieved by constantly obtaining feedback. This technical architecture is more and more common the further down, the more dedicated it is. More versatile, which means continuously lowering the threshold for the industry to use technology; More specialized, it is to go deep into the industry to deepen the application.
Therefore, Li Yanhong also emphasized the importance of basic software: "The chip card neck is very important, but the software card neck is also important." Software must be rooted in order for innovation to continue to occur and to make the business at the top more prosperous. ”
Li Yanhong judged that the commercialization of artificial intelligence will need to be groped in the dark for a while. "Because the digital transformation of many areas of the real economy has not yet been completed, and digitalization itself has not brought significant improvements in efficiency." The widespread penetration of intelligence will take time, and the huge lifting effect of intelligence on the real economy has not yet become a broad consensus. Li Yanhong said.
At the 2021 Baidu Create AI Developer Conference, Robin Li said that in the next five years, Baidu will cultivate 5 million AI talents for the society. Up to now, Li said that Baidu has cultivated more than 3 million AI talents for the society.
At the meeting, Baidu released a series of new scientific and technological achievements. For example, "clairvoyance" and "tailwind ear" in the field of smart travel: the industry's first technology to realize echo cancellation in pure software solutions on mobile phones, and BEV autonomous driving perception solution that integrates vehicle and road; The solution in the metaverse field is MetaStack, which can compress the development cycle from half a year to a year to as fast as 40 days.
Baidu has not yet released its fourth-quarter financial report, and from its third-quarter performance, Baidu achieved revenue of 32.54 billion yuan, a year-on-year increase of 2%, and is expected to be 31.77 billion yuan. Baidu Core (a combination of search services and transaction services) revenue was 25.2 billion yuan, up 2% year-on-year. Among them, online marketing revenue was 18.7 billion yuan, down 4% year-on-year and up 10% month-on-month.
As of the close of trading on January 10, Baidu's Hong Kong shares fell 0.61% to HK$130.600 per share, with a total market value of HK$361 billion.