SIIA has done many artificial intelligence (AI) spotlights this year where we have featured impressive, boundary-breaking technology in the space. We have also released a handful of issue briefs, culminating in the most recent brief that we’ve released on Algorithms and Ethics. What we have not done until this point, is feature how different countries and regions across the globe are prepared to handle AI, will benefit from AI, and how they plan to use AI in the future. In an effort to compare these regions to each other, we will begin publishing the AI Landscape series as an accessory to the AI Spotlight series where we will do just these things. We begin this series with a feature on China.
According to PricewaterhouseCoopers (PWC), global GDP will receive a boost by $16 trillion by 2030 as a result of AI technology. Nearly half of all that growth will come from China with AI increasing GDP in China by an estimated 1% each year. China understands the potential it has to be a global juggernaut in the development of AI technology.
Recently, China released its July 2017 “Next Generation Artificial Intelligence Development Plan” showing that China has a serious desire for a transformation into an AI powerhouse. The plan shows that developing and incorporating AI systems will play a key role in China’s economic future. China plans on increasing investment AI research and development, promoting public-private partnerships, focusing on career and technical education to form a pipeline of talent, and creating a regulatory body that will examine risks and disruptions.
China seems well-poised to accomplish many of these objectives. With over 1.4 billion citizens, it has the largest population in the world. As a result, it has the most users that are connected to the internet, about 730 million people, and thus generate data. To have success in the development of AI, it is necessary to have incredible sums of data. Here, China has an advantage. China’s population generates more data than almost all other nations combined.
Human capital and knowledge are also essential to have success facilitating growth for AI. China already publishes more research papers on AI and deep-learning than the U.S., though the U.S. still dominates in number of patent applications for AI.
Additionally, China’s gubernatorial structure also gives it an advantage. Chinese consumers are willing to provide information to platform companies such as Alibaba, Tencent and Baidu in return for services they value highly, and Chinese data protections laws accommodate this consumer willingness to share information without unnecessary procedural burdens. This governance structure gives Chinese companies the ability to collect and process large amounts of consumer information.
Data and Research
Machine learning algorithms are most effective when they can utilize an incredibly large data set. As mentioned previously, because China has such a large market and 730 million internet users, it has access to more data than any other country just by using its own data. According to the Wuzhen Institute, while the $2.6 billion in funding for Chinese AI firms is still well below the $17.9 billion in funding for U.S. AI firms, China’s funding is increasing fast. China’s three largest companies, Baidu, Alibaba, and Tencent (also known as BAT), along with smaller Chinese firms are rushing to build data centers across the country to house and analyze all of this data. Alibaba is planning to spend $15 billion on AI. Baidu increased its AI spending to $464 million from April 2017 to Jun 2017, 28 percent more than it did over the same period in 2016, with much of its resources directed towards autonomous vehicles. Tencent has not yet invested heavily in AI, though it has more data than Baidu and Alibaba and is thus well poised to become a serious AI competitor.
China is even using simulations to generate more data. Firms in China have used video games and other simulations to generate new data that may not otherwise exist for many different applications such as development of autonomous cars. Additionally, this is done to theoretically create newer, smarter algorithms with fewer data and people to generate it.
To further add to how successful China is at generating data, China has a high population of citizens that uses speech-to-text apps to send texts or write messages since typing characters can be tedious. This high utilization rate means that China has an overwhelming quantity of voice recordings which can be used for voice recognition in AI software for higher accuracy.
Finally, in terms of patents, the U.S. dominates in absolute numbers of total patent submissions with roughly 15 thousand in 2014. Comparatively, China had roughly 9 thousand. However, the increase in the number of AI-related patent applications from 2005 to 2014 was by 125 percent in the U.S. versus 300 percent in China. Additionally, in 2016, China surpassed the U.S. in the number of journal articles published on deep-learning. This is due to China being estimated to have over two-thirds of the world’s AI scientists.
Talent and Capability
Despite the increased focus on AI technology by China, the U.S. is still the leader in terms of talent when it comes to AI technology. China cannot utilize the data and innovate to realize the full potential of AI without access to some sort of talent pool. Part of its “Next Generation Artificial Intelligence Plan” places a large focus on STEM and Career and Technical education to fill this shortage.
Aside from AI scientists, other AI-related technology roles remain largely unfilled. By the numbers, China has 50,000 workers in AI-related tech jobs while the U.S. has 850,000. China’s greater focus on training and education in this space can potentially fix this problem. About 8 million students will graduate from Chinese higher education programs in 2017; and its growing. In the US we produce around three million and it’s been flat since the 1980s.
China also has strong skills in math, language, and translation that present it with the building blocks of success in AI capability. Chinese universities are beginning to lay a major focus on AI technology as well. Tsinghua University in Beijing announced plans to create a “military-civil fusion lab” to find multiple applications of AI, also fitting the multi-faceted goals of China’s AI Plan. Baidu is partnering with the Chinese Academy of the Sciences, Tsinghua University, and Beihang University creating China’s first national deep-learning laboratory.
Additionally adding to China’s AI capabilities, China is by far the biggest consumer of new installations of industrial robots in the world. Korea, Japan, Germany and the US have a head-start so they have installed more per worker than China has. But China is catching up fast.
Data Protectionism Poses a Challenge
Aside from the challenges mentioned earlier such as the creation of a talent pipeline, investment, and research and development, a major challenge to China in reaching its own full potential in AI has to do with its data protectionism policies. Much like other internet and data based technologies, for AI to succeed, data has to flow. Impediments to data flows, such as data localization requirements that China has supported, limit the full spectrum of data that China is able to access. For example, the PRC Telecommunications Regulation of 2000 requires all data collected inside China to be stored on Chinese servers. This regulation places a barrier for Chinese companies who could see costs rise with the mandate to store their data in China and deters foreign companies from operating in China.
China views these mandates as bolstering their own industries, as well as providing itself with greater control over the data that moves within China. However, this presents challenges to many of its own smaller businesses that may in some cases be able to store data outside of China for far less cost. ECIPE released a report that showed that data localization measures reduced the GDP of China by 1.1 percent. The economic losses of Chinese citizens due to such measures amounted to $63 billion. Needless to say, the AI ambitions of China may be at odds with its policies on data protection as it places limits on the data pooled for algorithms to be the most efficient.
China’s AI plan shows that it is willing to dedicate significant resources to developing its AI capabilities as part of a broad strategy. It has more data from itself than every other country’s data combined and the most internet users to continue to generate more data. It is working to produce more workers who specialize in AI technology related work. Data localization policies may offset the gains made to China’s GDP from AI. It also needs to amplify its talent pipeline. Until China fixes its labor shortage and revisits policies that hinder innovation, it will be difficult to rapidly overtake the U.S. as the global leader in AI. While China has some ways to go before it becomes the global powerhouse for AI that it so desires, there are certainly many factors that put it in a good position to achieve its goals.