BEIJING, CHINA – OCTOBER 13: A self-driving robotaxi drives along a road on October 13, 2020 … [+]
China and the United States are on par in AI development, but China’s implementation of the technology in products and services could take the lead by 2023.
As Kai-Fu Lee, an astute observer of China’s AI development, puts it, “We are now in the era of AI implementation.” While the West, especially the US and Canada, will maintain a lead in AI research, these Western advances are coming quickly Adopted by China, a huge market, a surplus of young engineers, government support and a cutthroat entrepreneurial culture are driving industrial innovation in artificial intelligence.
“The integration of the digital economy and the real economy is accelerating,” said Wang Haifeng, Baidu’s chief technology officer and head of Baidu Research.
China’s 14th Five-Year Plan and Vision 2030 both place great emphasis on the development of the digital economy, seeing this field as a source of enormous untapped innovation drive and room for growth.
“As pre-trained AI large models become the cornerstone of new technology, the level of intelligence has improved significantly, showing a larger space for large-scale applications from industry to scientific research,” Baidu’s Wang said.
Today, intelligent computing centers, deep learning platforms, and artificial intelligence models have formed a new artificial intelligence infrastructure in China, providing the foundation for the digital transformation of the real economy, especially manufacturing and autonomous vehicles.
In the next few years, the construction of new artificial intelligence infrastructure will surely become one of the important pillars of the development of the digital economy, become the investment focus of local governments in China, and contribute to regional economic development. If the AI infrastructure is expanded, we can expect to see China develop new products and new business models with industrial applicability.
Still, China lacks the academic infrastructure to drive U.S. technology. While China has established some powerful AI research institutes, such as the Beijing To hit hard – is decades away from the West. Chinese teaching and research cannot compete with private sector salaries, and universities cannot attract top academic talent from abroad.
However, thanks to the expansion of large-scale deep learning models, cross-language, cross-task, and cross-modality are becoming the main trends in the development of artificial intelligence in China today. According to Chinese artificial intelligence giant Baidu, “industry big models” are being applied in aerospace, finance and energy. Building artificial intelligence infrastructure around the needs of various industries, China is forming an “artificial intelligence + industry” pattern. Baidu is looking forward to the emergence of a wider industry model ecology, which will drive the intelligent upgrade of AI-driven in thousands of industries.
WUZHEN, CHINA – OCTOBER 21: Baidu CTO Wang Haifeng speaks at the 6th Congress … [+]
“As pre-trained AI large models become the cornerstone of new technology, the level of intelligence has improved significantly, showing a larger space for large-scale applications from industry to scientific research,” Baidu’s Wang said.
Baidu, one of China’s AI giants alongside Alibaba and Tencent, operates ERNIE-ViLG 2.0, a 24 billion parameter model trained on 145 million Chinese image-text pairs.
Another large-scale Chinese model is Taiyi, from the IDEA research lab led by Microsoft Research Asia co-founder Harry Shum, Microsoft’s largest research facility outside the US Taiyi was trained on 20 million filtered Chinese image-text pairs, and billion parameters.
Meanwhile, BAAI launched the latest version of Wudao in 2022, a pre-trained deep learning model with up to 1.75 trillion parameters.
The maturity of these key technologies, coupled with big data and cloud computing, has injected new impetus into the development of automation. Various types of artificial intelligence robots have shown great potential in real-time perception, intelligent decision-making and optimized control. The increasing maturity of AI-enabled robot technology means that they can be increasingly used in construction, mining, disaster relief and other work scenarios that require a lot of manpower.
With the entry of autonomous driving into the urban environment, the difficulty of sensing complex environments and the difficulty of processing massive data are greatly increasing. Traditional small models can no longer meet the requirements of high-level automatic driving, and AI large models have become the latest focus of automatic driving technology.
This AI large model technology enables self-driving cars to effectively expand semantic recognition data and exponentially improve the efficiency of solving long-tail problems. The rapid advancement of technology means that by 2023, the scale of commercialization of autonomous driving and the size of fleets in major cities in China will increase simultaneously, and the market penetration rate of autonomous driving smart cars will reach a new level. The smart car industry will move from the experimental stage to full maturity.
In 2023, the scale of commercial operation of autonomous driving and the size of fleets in major cities in China will grow simultaneously.
Baidu said that the deepening of data marketization has highlighted the urgent need for “horizontal and vertical” trusted data circulation networks. The development of such networks will continue in 2023.
Technology companies and scientists are also actively exploring “explainable” artificial intelligence technology, aiming to achieve more effective human-computer communication in the context of value alignment, so that artificial intelligence can truly understand human intentions and achieve more predictable artificial intelligence governance. In a more intelligent and digital society, possessing manageable and controllable artificial intelligence technology capabilities will become a new competitive advantage for enterprises.
Emerging digital infrastructures and tools allow for the creation of content-rich virtual worlds that have evolved from parallel to the physical world to increasingly merge with it.
Virtual simulation and artificial intelligence technology will create more accurate and smarter digital twins in shopping, manufacturing, leisure and other fields.
AI-generated content will enable entirely new modes of content creation for drawing, painting, and other media. Combining the immersive experience services brought by VR/AR and the high-speed data transmission capabilities provided by 5G, the extensive and rich metaverse industry is expected to accelerate its implementation and carry a new wave of products and service models. disruptive innovation.
In 2022, China released the landmark “Opinions on Strengthening the Ethical Governance of Science and Technology”, and submitted the “Position Paper on Strengthening the Ethical Governance of Artificial Intelligence” to the United Nations, calling for the principle of “people-oriented, intelligence for good” to ensure that artificial intelligence Safe, reliable and controllable.