Illustration: Peter C. Espina/GT
Recently, China’s artificial intelligence (AI) development has attracted domestic and international attention. The State Council issued a new plan to guide the development of China’s AI from basic technology to practical applications. The British Economist magazine wrote that because of talent and data advantages, as well as the strong support of the government, China may catch up with the US in the field of AI. As an AI practitioner, I think it is possible that the China’s AI technology can occupy the leading position in the world. The field will eventually narrow to a competition between US and Chinese technology.
Compared to many other countries, China has great advantages in AI development. The progress of AI mainly depends on three elements: algorithms, data and computing platforms.
The computing platform is the foundation everything else is built on. Currently, China’s cloud computing companies have already built an infrastructure that covers the globe with an ability to support any AI applications.
Algorithms need to be nourished by data. The huge economy and population provides more abundant data than other international competitors.
Nevertheless, if China wants to become an international leader in AI, it needs to step up efforts in many aspects.
First, China AI researchers should develop technology tightly bound with real life application scenarios. “Application scenarios” and “vertical integration” are the two most common jargon terms in AI, because thousands of AI studies are too abstract to be applied to daily life. Our AI technology must be bound with real-life scenarios to create actual social value, rather than abstract academic research. A platform aiming to be applied to all application scenarios will be the least useful.
In fact, Chinese AI workers have been diligently working toward breakthroughs in scenario-specific applications. For example, the “ET industrial brain,” an AI system designed by Alibaba Cloud Computing, a company affiliated with Alibaba Group to help manufacturers improve production efficiency, increased the ratio of error-free production by one percentage point in a workshop of photovoltaic product manufacturer GCL-Poly. This saved over 100 million yuan ($14.9 million) in production costs per year.
Second, China needs to create its own original basic technology to power AI, and tear down its reliance on imported ideas. Compared to other countries, China lacks a big picture view, technical capacity and reserves of talented workers. China’s AI industry is not in need of development environments but original algorithms and theory. Why were the theories of deep learning and neural networks not created in China? Do we have a chance to break new ground in the next round of the AI revolution? Solutions as well as in-depth thinking to these problems are needed for China to develop world class AI technology.
Numerous cases have increased our confidence. For example, Hangzhou’s central city computer can be called the world’s largest AI system. It can conduct a real-time analysis of data from tens of thousands of road cameras and make complicated decisions. This project has also spawned many other technical achievements, such as major developments of the machine vision algorithm evaluation platform KITTI, and three papers selected by publisher and conference organizer Association for Computing Machinery (ACM). In fact, the challenges of these projects are all world-class, but most of the solutions are coming from original China AI technology.
Finally, we should say no to the craving for short-term financial returns. We need to realize that the development of AI in China will create a high-tech future for the country and even the whole world. AI development requires an entire environment that supports science and technology. Any overeager dreams about quickly developing AI and desires for a quick return on investment will possibly hamper rather than help progress.
Compared with the US, Chinese domestic investors pay more attention to short-term financial returns, while US companies are investing in the future. Maybe some of the American projects will not earn back their investments immediately, but the groundwork is laid for the future. No great company or industry is built overnight. The well-known technology company Amazon experienced losses for 20 years before it started to earn profit. We have to look at the development of AI in with a vision beyond financial returns