China and Artificial Intelligence
While previous platform shifts have historically been driven by the western players, China has earned itself the front-row seat this time as a key player in the leap towards Artificial Intelligence (AI). Mckinsey has acknowledged this transition by naming China as “one of the global leading hubs for AI development”, and a “world leader in AI development” alongside the United States.
And the leading Chinese companies have proven that the excitement in them is more than hype. For instance, Baidu managed to attain the milestone of surpassing human-level language recognition performance in 2015, almost a year before Microsoft’s announcement of a similar feat in October 2016. Baidu’s feat is arguably more impressive given the complicated lingua landscapes in China as a result of Chinese dialect-influenced tonal differences for spoken Mandarin.
Truth of the matter is, this ascension to the forefront of cutting edge technology development is not a fluke. China has been working hard to restore itself to global prominence in Science and Technology, with sustained and committed investment in education, training and research over the last few decades. Seen as the next major technological leap that will have profound implications for the economy and society, AI is naturally the key priority and has received overwhelming support and acceptance in China on all fronts.
Part one of this two-part essay aims to provide an overview to the development of AI in China.
Within the political circle, AI has been elevated to the status of strategic importance by the technocratic government. The term “Artificial Intelligence” first entered the lexicon of the government’s external communication in 2016 in President Xi’s address to the G20 delegation at Hangzhou. This was followed by a working report address by Premier Li Keqiang at the “Two Session”, which places AI at the forefront of China’s technological focus areas despite its debut inclusion in the report. This is an important sign post to the leadership’s thinking and China’s policy direction.
 The Two Session is a key gathering of political representatives and leaders in China where work achievement is reviewed and objectives/ policies direction are set.
Strategically important emerging industries, including new-generation information technology such as Artificial Intelligence are to account for 15% of GDP by 2020.
— China’s State Council
This policy clarity helps to set in motion the various governmental machinery, in support of AI development as stated by the leadership. The result was a 3-year action plan for AI released in 2016, spelling out tactical responses to research, funding, talent, intellectual property, international cooperation, industry development and standards setting.
A result of this effort was investment by the central government into “Key State Labs” with Baidu, Alibaba and Tencent. Designated state labs will receive resources in return for working on government assigned projects or objectives. Baidu’s lab ( Key State Lab for Artificial Intelligence with Baidu ; 深度学习（AI）国家工程实验室), for instance, will be partly run by academics associated with Chinese military projects in the past.
The local governments, which compete among themselves within China, have not stood still. The front runners have embarked on early industry development efforts and backing up their commitment with resources. Suzhou City, which lies north-west to Shanghai, has reportedly (by New York Times) dangled support of up to US$800,000 for leading AI companies while AI entrepreneurs could look forward to a cool US$1 million support from the Shenzhen government.
Speed of response aside, the action plan was also notable for being jointly issued by NDRC, MIIT, MOS, MOF and CAC. Each of these agencies are known to be powerful and notoriously territorial in China. The successful joint statement was thus also an indication of political will and a strong statement of intent.
 National Development and Reform Commission, Ministry of Industry and Information Technology, Ministry of Science, Ministry of Finance, Cyberspace Administration of China
AI is expected to generate significant economic spin-offs for China. Mckinsey Global Institute have projected a gain of 0.8–1.4 percentage points (depending on speed of adoption) to China’s GDP growth. The estimate takes into consideration only impact from AI-led automation.
AI ERA, a China-based technology consultancy, has projected the market size for AI to be at more than RMB 20 billion in 2017 (CAGR of >40% from a base of RMB 6.9 billion in 2015), and market size for related industries at RMB 100 billion. Notably, China’s AI application market is projected to grow 50% annually in the coming years, outstripping the 20% projected compounded annual growth rate in the global market.
Similar to the United States, start-ups contribute significantly to the base of activities in the AI space. RMB 9 billion(~US$1.3 billion) was raised by AI Start-ups in China in the 12 months leading to October 31 of 2016, more than the aggregate investment and deal volume for all previous years. In comparison, CB Insights reported a global funding of US$5 billion into AI-related start-ups globally for the whole of 2016. Further to activating the AI start-up scene in China, Chinese money has also been pouring into US’s AI start-up in search of cutting edge technology, to the extent that it has become a national security concern for the Pentagon.
Given the data-centric nature of the latest dominant AI techniques, hyper-scale computing players are among the most active globally in driving investment and activities in AI. This phenomenon is also observed in China, with Baidu, Alibaba and Tencent (BATs) leading the pack for AI advancement. The 3 companies are among the leading investors into R&D in the Silicon Valley. These have helped the three technology giants to become credible competitors for leading international talent in the AI Space increasingly. The overview of BAT’s venture into AI is illustrated below.
China has always been receptive of new technology. We have witnessed this phenomenon previously, with the successful proliferation of mobile computing, 4G and cashless payment. AI has been embraced in the same manner.
The embrace of “new technology as a pathway to progression” is a social narrative that is deeply entwined into the social fabric of Chinese psyche by 70 years of rule under the Chinese Communist Party. Unsurprisingly, public opinion surveys conducted have indicated that most hold favourable opinion of AIs (61%), are concerned about the country’s progress in AI development (67.9%), and support further proliferation on training and education for AI (77.8%).
Note: This receptivity by the general public is perhaps the one factor that stands China out from other countries around the world when it comes to AI. I will talk about this in greater depth in Part Two of this essay.
Academic Strengths and Weaknesses
Notable world-class researchers of Chinese origins have been prominently represented in the AI academia community. Many of them have risen through the Chinese education system, did their post-graduate studies in the United States before achieving prominence globally for their work while based outsides the United States. With the spotlight falling on AI, these researchers have become prime hiring targets for Chinese corporate and universities. The capture of such “superstar researcher” has provided a short-cut to “turbo-charging” the Chinese AI research, its talent development and the overall ecosystem.
International bench-marking competition has also been dominated by Chinese-based teams. The famed ImageNet challenge is known for bringing the public spotlight onto Deep Learning. In the 2016 edition of the challenge, Chinese-based teams dominated the various challenge categories. Winning teams ranges from university research outfits to private companies and even public security agency-affiliated research groups. Similarly, speech recognition company iFlytek has also begun to dominate United States’ National Institute of Standards and Technology bench-marking exercise.
Despite these encouraging signs, the average standard of AI research still lags that of the US (though the situation could be rapidly improving). While China accounts for the largest share of research publication (almost 2x that of US at No. 2), it only ranked at a measly 34 based on Field-Weighted Citation Impact (compared to No. 4 for US). This suggests that the quality of research coming from China have not been strong despite the quantity.
AI being an overwhelmingly talent-reliant industry, China’s biggest challenges could come from the readiness of its own talent ecosystem to produce the required people to sustain the growth of AI in the country. It is interesting to note that almost all of the globally prominent ethnically Chinese AI researchers did their career defining work in the United States.
“弯道超车” (Wan Dao Chao Che) is a common saying in China. Literally, it meant “overtaking at the bend” and is thrown around plenty by CEOs during interviews when asked about transformation and competitions.
In many ways, “over taking at the bend” is an very apt description for what is happening in China. China recognizes it has been a follower in the race for economical and technological dominance. This impending platform shift is going to be the bend that is going to slow everybody down. But it is also going to be an opportunity for China. While she might not have the best car in the circuit at the moment, she is throwing everything she got into preparing herself for the overtaking at this next bend. Watch out!