Most Viewed Content:

Ford will launch a new electric truck under a new product line

REcent report prove that Ford is planning on launching a new electric vehicle that will come with a new nameplate.

Samsung ships first 3nm Gate-All-Around chips to cryptocurrency mining industry customers as promised

This Monday, July 25, Korean electronics giant Samsung began...

Xiaomi 12S and 12S Pro Revealed With A New Processor

Xiaomi is now on a device launching spree as it is set to launch the Xiaomi 12S and 12s Pro with a new Snapdragon processor.

After L2, the first commercial use will be L4 rather than L3

2022 World Artificial Intelligence Conference opening ceremony, Baidu founder, chairman and CEO Robin Li speech pointed out that in the past year, both in the technical level and in the commercial application level, artificial intelligence has made great progress, some even directional changes. When it comes to autonomous driving, Robin Li pointed out that the first to enter commercial use after L2 is likely to be L4, not L3.

This year, milestones in the field of autonomous driving in China have been constantly refreshed, and various related policies and regulations have been introduced like a spring, providing an internationally leading policy environment for commercialization and large-scale expansion. As of July, the cumulative order volume of Baidu Radish Express self-driving travel service exceeded 1 million orders, operating in more than a dozen cities including Beijing and Shanghai. Chongqing and Wuhan were opened for fully unmanned commercial operation respectively.

The first to enter commercial use after L2 is likely to be L4, not L3, because the responsibility for accidents is clearly defined for both L2 and L4. L2 responsibility lies with the driver, and L4 operators are responsible for accidents. l3 is different, as the driver takes over when needed, which makes it difficult to define responsibility for accidents. Therefore, he believes it will take longer for L3 to become widespread.

In addition, technological advances have increased the ability of autonomous driving to generalize, and the scale effect is gradually emerging. “When we want to be qualified to operate autonomous driving in a certain area of a city, technically it generally only takes about 20 days of preparation time because the generalization of the technology is already good and our autonomous driving is not achieved by over-fitting a specific area.” Robin Li said. (Wenmeng)

With the full text of Robin Li’s speech.

Dear leaders and guests, good day!

It is a great pleasure to come back to Shanghai to participate in the 2022 WAIC World Artificial Intelligence Conference, which has been held for four consecutive years, and its global influence and “gravitational field effect” are increasing, the scale of Shanghai’s AI industry has multiplied, and the construction of world-class industrial clusters has taken solid steps. The new conference will help Shanghai AI development to achieve a new leap forward.

In the past year, both in the technical level and in the commercial application level, artificial intelligence has made great progress, some even directional changes.

The AI painting that you just saw is a representative of the progress at the technical level in the past year. The reason why we say there is a directional change is that AI has moved from understanding language, text, pictures and videos to generating content, and Higaga’s AI painting automatically generates pictures of various styles through text descriptions. This is an example of the automatic generation of article stories through text description of the topic. Some video content in Baidu APP today is the result of AI automatically converting the graphic content of Bajia into video. These are all AIGC, or artificial intelligence automatically generated content.

The technology behind AIGC is the so-called pre-trained big model, and many of us here are technical experts in AI, so I believe we will cover this technology many times in the subsequent speeches. What I want to say is that AIGC will disrupt the existing content production model, and can realize the creation of content with unique value and independent perspective at one tenth of the cost and a hundred times and a thousand times the production speed.

Of course, what is more exciting is the progress at the commercial application level. Artificial intelligence has been on fire for so many years, but business should always be one of the weaknesses, and the lack of good business prospects will make startups stagnant growth, huge losses, and difficulties in financing and listing, while large companies will become increasingly ungrounded, either gradually becoming pure research departments or gradually becoming an appendage of other businesses.

When it comes to commercial applications, the most obvious progress is still in the field of autonomous driving. In June this year, GM-backed Cruise opened a fully unmanned self-driving commercial operation in San Francisco, and although there were various stumbles in the middle, they persevered and are expanding the scope of operation. In China, Baidu’s Radish Run accumulated more than 1 million orders in July, operating in more than 10 cities including Beijing and Shanghai. Earlier this month, Chongqing and Wuhan opened their fully unmanned commercial operations for Radish Run respectively, providing an internationally leading policy environment for the commercialization and scale expansion of driverless vehicles in China.

In my opinion, there is also a change of direction involved here. It used to be thought that driverlessness was still a long way off, and even Turing Award winner Sfarski believes that it may take decades to achieve complete driverlessness. The first thing to enter commercial use after L2 is likely to be L4, not L3, because the definition of liability for accidents in L2 and L4 is clear. The difference between L4 and L5 is that L4 is a limited range driverless, and L5 is an unrestricted range driverless. The difference between L4 and L5 is that L4 is a limited-range driverless vehicle and L5 is a non-limited-range driverless vehicle.

In addition, from our practice, the speed of technological progress in autonomous driving is beyond expectation. When we want to be qualified to operate autonomous driving in a certain area of a city, technically it usually only takes about 20 days of preparation time because the technology is already very general and our autonomous driving is not achieved by transition fitting to a specific area.

Today, citizens in more than 10 cities can experience the Turnip Express self-driving service, and autonomous driving is very close to us. The public’s trust and welcome for autonomous driving is also increasing. A survey shows that 83% of Chinese people accept autonomous driving technology, and Chinese consumers have a higher demand for connected and intelligent cars, as well as a higher level of welcome and tolerance.

Of course, car manufacturers are also actively embracing autonomous driving. Many auto OEMs realize that doing self-driving R&D from scratch is neither economical nor efficient and is not competitive, and are more willing to work with us. At present, there are more than 30 domestic and foreign mainstream car manufacturers cooperating with Apollo. Baidu’s Jidu Automotive is also a partner of Apollo. In June this year, Jidu released its first robot concept car, robo-01, and the mass production model will be available in 2023. It is an intelligent car that can move freely, communicate naturally and grow up by itself, embodying the “smart awakening” of automobiles.

In addition to autonomous driving, we have seen progress in the commercialization of artificial intelligence in a number of areas over the past year. The most obvious is in the intelligent transformation of infrastructure.

The first is intelligent transportation. Currently, China’s road traffic network is not yet able to improve traffic efficiency and reduce accident rates through real-time signal regulation and vehicle-road coordination, and urban congestion makes many people waste a lot of time on the road. In order to relieve traffic congestion, localities have to implement policies to restrict the purchase and driving of cars, which curbs the consumer demand that should be there and does not fundamentally solve the problem. According to our practice in various places, the intelligent transformation of transportation networks can improve the efficiency of traffic by 15-30%, which means about 2.4%-4.8% annual growth of GDP. Currently, Baidu’s intelligent transportation solutions have been put into practice in more than 50 cities across China. Just a few days ago, the Ministry of Transportation officially listed Baidu as a pilot unit of the strong transportation country, carrying out pilot projects in high precision maps, smart cars, smart roads, cloud platforms, and intelligent transportation industry ecological development.

It can be foreseen that with the improvement of traffic efficiency, the policy of restricting the purchase of cars and traffic restrictions will go into history, injecting new vitality into the economic growth after the urban epidemic.

The second is the intelligence of energy and water infrastructure. China has established a strong physical network of infrastructure in the fields of energy, water conservancy, water supply and heating, but the past construction, with emphasis on hardware and light on software, was not very intelligent. This year, with the high temperature weather across the country and record high electricity loads, the entire grid system is stretched so tightly that even a small fault can easily lead to a massive blackout. Now, many of China’s provincial power grids are using Baidu Intelligent Cloud’s AI inspection, which can inspect 24/7, improving inspection efficiency by 6-10 times and effectively ensuring the safety of power supply. We believe that the next step should be to strengthen the top-level design of resource deployment in water and electricity systems, accelerate the intelligent transformation of these infrastructures, and use AI to achieve efficient real-time resource scheduling.

In addition, in the field of industrial Internet, with the unique advantage of cloud intelligence, Baidu Intelligent Cloud has created an AI+industrial Internet platform “Kaiwu”, which has been selected as one of the national “Double Cross Platform”. Kaiwu is helping Chinese enterprises reduce costs and increase efficiency in key scenarios such as quality management, safety production, energy consumption optimization, logistics scheduling, etc., and improve innovation capabilities, helping China transform from a “manufacturing power” to a “manufacturing power”. For example, in quality management, it only takes one second for a car factory to complete 22 points of quality inspection of headlights; in energy consumption optimization, we use AI to help a thermal power plant optimize the energy consumption of air-cooled island equipment, achieving a reduction of 1.55 grams of standard coal per kilowatt-hour of electricity. Based on 1000 air-cooled units nationwide, the carbon reduction potential can reach 6 million tons per year, helping the country to achieve the “double carbon” target.

The commercialization of AI in these fields requires end-to-end technology tuning for each industry. Baidu has been in the field of artificial intelligence for 10 years. In these 10 years, we have invested more than 100 billion in research and development, and each year the proportion of research and development is more than 15%, last year it reached 23%, which is rare in the world’s large technology Internet companies. This kind of pressurized, marathon investment, so that we have leading self-research technology in all levels of artificial intelligence, from the bottom high-end chip Kunlun, to the flying paddle deep learning framework, and then pre-trained large models, (we recently launched a large model of the financial, electric power, aerospace and other industries) to finally achieve a substantial increase in the application of efficiency.

Of course, we are also aware that the digital transformation of many areas of the real economy has not yet been completed, and digitalization itself has not been able to bring about significant efficiency improvements, the penetration of intelligence still needs time, and the huge role of intelligence in boosting the real economy has not yet become a broad consensus. Therefore, the commercialization of artificial intelligence still needs to be groped in the dark for some time. But a new thing, from “no one good” to “no one can reach”, the decisive victory is often in the word “persistence”. This is especially true for science and technology innovation.

Science and technology innovation is inseparable from the supporting system innovation. The need for greater reform and innovation efforts to give innovation the best development environment. For example, at present, the popularity of unmanned vehicles still face “four not a difficult” policy barriers, that is, unmanned vehicles can not enter the market, can not be licensed, can not remove the safety officer, can not operate the charge, accident liability is difficult to identify. China’s autonomous driving technology is at the forefront of the world, but the opportunity is also fleeting, need to promote institutional innovation and further breakthrough policy bottlenecks. Only in this way can the two-way run of artificial intelligence and the real economy be realized and the great progress of society be promoted.

Finally, I wish this Shanghai Artificial Intelligence Conference a great success! Thank you all!

Latest

Classic FPS game Thor’s Hammer was ported to Apple Watch

A programmer managed to get the original "Quake" (Thor's...

Ring Finally Brings End-to-End Encryption to Its Video Doorbell

More than a year after adding end-to-end encryption options...

Nvidia Geforce: The final clearance RTX 3060 has three new products

RTX 40 series is coming soon, but what I...

Android 13 is making it easier to separate work and personal data

Android 13 has some new features designed to improve...

Newsletter

spot_img

Don't miss

Classic FPS game Thor’s Hammer was ported to Apple Watch

A programmer managed to get the original "Quake" (Thor's...

Ring Finally Brings End-to-End Encryption to Its Video Doorbell

More than a year after adding end-to-end encryption options...

Nvidia Geforce: The final clearance RTX 3060 has three new products

RTX 40 series is coming soon, but what I...

Android 13 is making it easier to separate work and personal data

Android 13 has some new features designed to improve...

Tencent plans to increase holdings in French veteran Ubisoft

After investing in Japanese game company FromSoftware last week...
Threza Gabriel
Threza Gabrielhttps://www.techgoing.com
TechGoing is a global tech media to brings you the latest technology stories, including smartphones, electric vehicles, smart home devices, gaming, wearable gadgets, and all tech trending.
spot_imgspot_img

August Steam hardware and software report, Linux system accounted for another increase

According to Valve's latest Steam hardware and software report for August, Linux devices have increased slightly as a percentage of the overall market share....

1.5K high-frequency dimming flexible screen will cover the 1.5K-2K price range

After Xiaomi Redmi K50 Supreme Edition was released on August 12, using 1.5K OLED straight screen, Tianma and Huaxing photoelectric two screen partners jointly...

Tesla will soon enter the Thai market and start hiring consultants

In early August, Tesla Shanghai super factory completed the production line upgrade, and Tesla's electric car production speed greatly improved. With the increase in...