In 2024, if someone hasn’t heard of artificial intelligence, they are definitely not in the loop. AI, short for Artificial Intelligence, is widely known and ever-present. From the Granolas composed of 7 tech giants in the US to the leading Al sector in the Shanghai and Shenzhen stock markets in China, and NVIDIA reaching a market value of two trillion dollars with a few AI training chips, the AI technology is singing triumphantly in the background. Sora is quickly replacing the metaverse, shaping people’s perceptions and influencing the unpredictable changes in the financial market. All institutions and individual investors are holding their breath, watching closely: How much more can Al rise? In parallel, with the support of AI technology, autonomous driving is making its way into our lives. Can it become a common scene for human transportation? In fact, there are more serious and practical questions facing us. Question one: Where are the technological barriers and legal risks of autonomous driving?
When considering the development of artificial intelligence in the automotive industry, a symbolic representative of the era of explosive growth is NVIDIA. Last month, NVIDIA, the global chip king, released a new chip processor: Blackwell GPU. NVIDIA CEO Huang Renxun claimed that the GPU has 208 billion transistors, compared to the 800 billion transistors in last year’s H100. Huang also assured that the chip’s performance in training AI models has doubled compared to the current generation of graphics processing units, and its “inference” capabilities have increased fivefold. At the same time, the development of artificial intelligence technology has brought about rapid iteration and popularization of autonomous driving technology. The penetration rate of smart driving cars in China is increasing year by year. According to Huawei’s prediction, by 2030, the penetration rate of new autonomous driving vehicles in China will reach 20%, the entire vehicle computing power will exceed 5000 TOPS, and the penetration rate of C-V2X will reach 60%. However, is autonomous driving technology really safer than human driving, as some experts claim? First, let’s look at radar. Radar theoretically cannot cover all scenarios perfectly. Although the probability of autonomous driving cars stalling or shutting down when faced with temporary road construction and roadblock cones is decreasing, cases of test vehicles being involved in traffic accidents are still common. Last year in the United States, General Motors was issued a driving ban by the court due to a RoboTaxi taxi hitting a pedestrian, leading to the suspension of production of that model. Currently, in China, in first-tier cities and airports where autonomous driving taxis are permitted to operate, human assistance drivers are still required to be equipped in the driver’s seat. Although their hands and feet do not need to be in action, when emergencies occur, the display screen will immediately issue an alert requiring the driver to regain control of the vehicle. In fact, vehicles on the market equipped with L2-level autonomous driving technology already have this feature. Taking Mercedes-Benz as an example, the Mercedes-Benz GLE 2024 and the all-new E-Class, which were launched in July last year and March this year respectively, both support intelligent navigation limited to L2-level autonomous driving. However, the steering wheel does not support long periods of hands-free operation. The display screen usually flashes red images a few minutes after autonomous driving is activated to warn the driver to return their hands to the steering wheel. After driving for a while, it will return to the steering system with a green icon. The time for hands-free operation within the lane can gradually be extended through OTA upgrades in the future. When it comes to visual cameras, even the most advanced light-sensitive devices have imaging rates between tens of millions to billions of pixels, while the human eye has the ability to perceive images with 600-700 million pixels. Although Musk’s brain-computer interface can reflect image signals from the light-sensitive nerves in the brains of blind people, it cannot completely replace the clarity of the human eye, especially when it comes to the instantaneous demands of vehicles. For example, facing an unavoidable accident in extreme situations, but with a choice of collision objects, what should be done? When it comes to the typical human dilemma of “saving your wife or your mother first,” cameras may only choose the visual object based on the extent of vehicle damage. The touching story of the bus driver who chose to collide on his side to save all passengers on the bus is no longer possible under the control of electronic cameras. The debate between automatic driving cameras and radar has not ended because of the foolish answer of “I want both.” From minor technical obstacles like changes in LiDAR to major difficulties like complete scene coverage, there is no easy mode, and the problems are far from being resolved. On the contrary, if the technology is rapidly popularized and fully implemented, leading to a collapse and market exodus, the “Chernobyl moment” is looming. Therefore, what is the legal definition of automatic driving? This is an unavoidable issue in the process of promoting automatic driving. Representatives of the National People’s Congress in the automotive industry, such as Wang Fengying and Feng Xingya, have repeatedly emphasized and closely monitored the determination of responsibility and claims for traffic accidents under automatic driving technology in proposals last year and this year, hoping that legislative bodies will soon introduce relevant laws and regulations. At the beginning of this year, Lotus achieved an IPO listing in the United States through a “special purpose acquisition company” and soon announced that the manufacturer would intervene and assume partial compensation responsibility in the event of a traffic accident involving its autonomous driving vehicles. It should be acknowledged that we must affirm Lotus’s pioneering behavior, but the low sales volume and high value of Lotus indicate that not all competitors can emulate this move. Whether this observation object will encounter similar problems in the future, and how to compensate and take responsibility if it does, will surely provide us with new insights into the development path of autonomous driving technology. However, before that, the risks for followers will always be like the sword of Damocles hanging over their heads. Question 2: What is the marginal in economics for autonomous driving?
In the post-epidemic era and the economic black hole caused by geopolitical conflicts, academia is revisiting Hayek, re-reading Keynes, and seriously studying Gu Chaoming’s analysis of the Japanese economy in search of the possibility of a surge in the Japanese stock market and leading Japan out of the lost twenty years, and the theoretical basis behind it. Marginal is mentioned the most: the margin dominated by state intervention and industrial policy, the margin of the free market… When we open Keynes’s classic “General Theory of Employment, Interest and Money,” many of the terms about the margin are his creations, such as total demand margin, total supply margin, marginal efficiency, marginal cost, marginal function, and so on. For economic participants in new concepts such as new energy vehicles and autonomous driving, the margin is actually the invisible line between cost and commercial profit. In other words, it is the dividing line that can maximize your profit, and crossing it is not enough. And the marginal cost here is not just the narrow sense of “how much money I invest,” “what is the return on investment ROE,” “how many times is the stock price-earnings ratio safe.” Otherwise, why is NVIDIA’s 20 times P/E ratio considered healthy and controllable, while BYD and Ideal Auto’s dynamic P/E ratio of less than 10 times are considered unsafe. These are all visible variables. According to economics, the total factor cost should be the maximum benefit that other unselected options can bring when we choose autonomous driving and charge down this path without looking back, and the maximum cost we pay for a single choice. Once this margin is crossed, the meaning of the career we are fighting for will undergo a disruptive change. Is it too late to turn back at that time? Even purely in terms of returns, the impact of autonomous driving technology on corporate profitability is not to be underestimated. Earlier this year, Apple announced the termination and exit of the autonomous electric vehicle project, which is by no means a case of starting and abandoning. Behind it are profound considerations and choices for the future of technology and market changes, as well as a keen sense of revenue and profit. Last year, Apple’s EBITDA pre-tax profit margin was 34%, while Tesla and General Motors were 9%. With slowing sales growth, cutting costs is also a top priority for Apple. In China, BYD’s gross profit margin is as high as 22%, but the net profit after depreciation is only a little over 5%. Chairman Wang Chuanfu stated in his New Year speech that the key to smart car development is cost reduction. He is confident in reducing the installation cost of lidar from 3000 yuan (410$) to 900 yuan (120$). However, cost pressure and profit killers will always accompany BYD’s path towards intelligence, especially with the price war background. After Xiaomi’s high-profile entry into the market in early April, Xiaopeng and Weimar both responded with price cuts. There is no sign of the price war easing in the short term. If BYD chooses to continue to suppress competitors’ profits in the mid-to-high-end market, autonomous driving technology will get closer to Lei Jun’s bottom line. There is no conclusion yet, but based on Lei’s style, he is willing to admit at the press conference that “SU7 pricing is definitely losing money,” so the profit-killing power of autonomous driving is also absolutely within his expectations. If the input-output margin of autonomous driving technology is related to the profit and survival of the company, then the marginal return on capital development of autonomous driving technology is related to the safety of the automotive industry, even the future of manufacturing. Autonomous driving technology urgently needs top-level design and overall planning for regional development. According to the information statistics released in the work reports of the provincial-level people’s congress in 2023, 24 provinces have already made corresponding layouts for new energy vehicles, intelligent connected vehicles, and autonomous driving technology, some with large investments. Will these layouts conflict with each other, causing contradictions with the macro-control goal of a unified national market? Will it waste the total cost and benefit of the whole society? Based on past experience, rapid and repetitive construction and development often lead to varying degrees of resource waste. Will the current stage of new energy vehicles, temporarily caused by supply and demand contradictions, turn into long-term, structural, and difficult-to-digest excess production capacity? This is a very serious issue. If the overseas battle is under great pressure and restrictions in terms of time and space, then in such a closed-loop environment, the development of autonomous intelligent vehicles in the domestic market at all costs is not only a bright path, but also a rugged and uncertain future. These are not issues that can be replaced by emotions and courage, but issues that are worth our repeated consideration and serious thinking. Question Three: Should the technological philosophy and ethical controversy of autonomous driving be considered in the overall development strategy of AI? There is a famous joke among the people: Aliens observing Earth report that the masters of Earth are cars, which feed on gasoline, while humans are parasites living in cars, lying around all day without moving. Now, by changing gasoline to electricity and adding autonomous driving technology, we can turn the alien report into reality. In 2022, the boss of Ferrari lamented, “Although Ferrari is willing to use new energy and autonomous driving technology, as a brand that represents driving pleasure, he does not believe that humans will completely hand over their understanding, experience, and enjoyment of driving to machines and chips.” Some experts in the industry also reflect that the development of foreign autonomous driving technology has stagnated, the penetration rate of electric vehicles has slowed down or even decreased, which is caused by both passive factors such as industrial policy choices and geopolitical games, as well as objective impacts caused by consumer questioning, dealer losses, and numerous difficulties in actual use. These emotional reasons cannot be simply explained by arrogance, indifference, selfishness, and unwillingness to admit failure. Recently, our country has organized several international artificial intelligence seminars and issued a collective declaration and action guide for the government and scientists. These documents emphasize that artificial intelligence and its applications should be considered in the context of the overall impact on human society by AI development, and should be studied and discussed cautiously and seriously using a similar overall organizational model as climate change conferences. AI cannot replace human thought, let alone replace humans in running society. Similarly, humans cannot completely entrust their safety to AI autonomous driving. The development of autonomous driving technology cannot replace driving schools, much less replace the cultivation and protection of human judgment, thinking, and action. Last year, the global recruitment company Nash Squared conducted a survey of global technology industry leaders and published the results in foreign media. According to the organization’s estimates, in the next five years, “automation” technologies including artificial intelligence will lead to a loss of 14% of jobs in the manufacturing and automotive industries. Assembly line workers, quality control assessors, and machine operators are the most likely positions to be replaced by artificial intelligence. Gabriel Edel, head of Google Cloud Manufacturing, Industrial, and Automotive sectors in Germany, told British media that in the automotive manufacturing field, AI-driven machines and equipment can operate more accurately and consistently than human operators, meaning less human intervention is needed in the manufacturing process. Of course, the automotive industry can retrain excess workers through AI training. China’s concentration and efficiency in education and training are even higher than those of European and American countries. However, a necessary precondition for job transition is full employment. Currently, the excessively competitive domestic market has led to layoffs and production stoppages. According to the National Information Center, the number of employees in China’s automotive manufacturing industry has decreased by nearly 500,000 from its peak in 2018 to 2023. This indicates that the development of automotive new four modernizations does bring certain side effects. The entire industry should quickly curb this trend to avoid a decrease in employment opportunities, otherwise successfully transitioned workers will also struggle to find employment. China’s population size, ethnic traditions, and public morality have their own distinct characteristics. The socialist core values also advocate for citizens to work hard and strive. There is rarely any denial or belittlement of the subjective initiative of the working people in society, instead emphasizing the spread of the idea of praising machines replacing workers. Technological progress and hard work are not contradictory. We will not allow AI to replace all workers’ jobs, or even allow social governance to be handed over to AI. This is not possible and will not be allowed to happen. In short: artificial intelligence should not replace human work, and artificial intelligence should certainly not erode human intelligence. Returning to the automotive industry, even if one day truly achieves autonomous driving and becomes widely popular, the time freed from driving fatigue and traffic management must be utilized effectively. Referring to the proposals on autonomous driving from representatives and members of the automotive industry at this year’s two sessions, as well as recent statements and regulations from the National Development and Reform Commission, the Ministry of Industry and Information Technology, and the Ministry of Public Security, there is a suggestion: treat autonomous driving and assisted driving technologies differently, avoid or even restrict the promotion and development of fully autonomous driving technologies, prohibit the use of misleading terms such as “autonomous driving.” For example, accelerate the popularization and application of Level 2 assisted driving, rapidly accumulate big data; focus on expanding the commercial application scenarios for Level 3 autonomous driving, increasing application frequency. Currently, the approved areas for autonomous driving taxis in Beijing, Shanghai, and other places can be expanded, but there are few actual users, and the accumulation of experimental data is relatively lagging. Level 4 autonomous driving technology lacks widespread scenarios and necessary reasons for promotion from both an economic and ethical perspective, and should not be excessively supported for development in the short term. Additionally, in the expansion of overseas markets, autonomous driving technology can broaden its application scenarios, accumulate experience and data. The National Development and Reform Commission and the Ministry of Industry and Information Technology have been emphasizing the necessity of coordinated efforts in going global, requiring companies to strengthen foreign joint ventures and cooperation, make good use of free trade zones and international trade agreements, steadily develop automobile exports. Meanwhile, in light of the domestic competitive situation, various ministries also require a practical increase in the market access threshold for new energy vehicles, only encouraging high-quality enterprises like Xiaomi to develop efficiently to avoid rushing and low-level redundant construction, and prevent waste of resources. It is believed that once the top-level design is implemented effectively, with a coordinated arrangement and development strategy tailored to local conditions, the prospects for the healthy and harmonious development of autonomous driving are optimistic and bright. Autonomous vehicles in China are developing in stages and levels, contributing to the formation and growth of new productive forces in China, the development and security of the global technological revolution led by AI, and the construction and prosperity of the community of shared future for mankind.