China: Blazing the AI trail
The race to lead on Artificial Intelligence
20 February 2018
China is fast catching up with the US in the race to be the world leader in AI. In the last five years, China has made remarkable progress and is fast closing in on its main competitor as it leverages proactive government support, huge data sets, a fast-growing pool of talent and increasing investment.
While the tech sectors in western countries may have difficulty in launching new technology due to regulatory constraints, China has the advantage of unparalleled, activist governmental support. In China, the "launch now and approve later" attitude is allowing AI development to flourish.
Recent policies such as the milestone 'Made in China 2025' plan aimed at bolstering smart manufacturing, robotics and semiconductors; the 'Next Generation Artificial Intelligence Development Plan' which aims to make China the world's primary innovation hub by 2030, as well as other supporting action plans and policies, including a three-year action plan to foster the development of artificial intelligence from 2018 to 2020, demonstrate the country's seriousness in becoming a world leader in AI. The Chinese government is providing funding for these plans too. The Next Generation Artificial Intelligence Development Plan was backed by an investment promise of US$150 billion to fund AI over the next decade.
The Chinese government provides special programs for top AI companies and start-ups, including free rent, subsidies for local hiring, housing and private school for top talents to encourage innovation. As a result, China has a large number of start-ups valued at more than $1 billion.
Data has become the world's most valuable resource – "data is the new oil" we are told frequently. If this is the case, China is data rich. China currently generates almost as much data as the US and Europe combined. This provides China with a great opportunity to take the lead on AI, where often the size of available datasets is key. The more data is available, the faster the algorithms can learn and develop, which should result in smarter AI offerings.
But how does one country generate so much data?
Collecting data is comparatively easy in a country that has nearly 800 million internet users, the majority of whom may be less concerned with the use of their data by tech companies than is typically the case in the US or Europe. The relative lack of constraints on re-using that data is critical for China's success when compared against the data protection driven caution in the EU and US.
One example is Mobike, China's leading bike-sharing company and one of the world's largest internet of things networks which has an average of 30 million riders per day. The bikes use various sensors including, GPS, accelerometers and Bluetooth, to transmit approximately 30 terabytes of data daily to a cloud-based server. To put that in context, globally, Twitter generates approximately 12 terabytes of data daily.
One of the key strengths of the Chinese AI sector is the large- and rapidly growing- pool of tech talent. While China may have previously been lagging behind in terms of top AI talent and experts, this is now changing. Many Chinese universities have opened programmes on AI and China now has a steady flow of young scientists trained in AI and data science. The number of academic papers produced by China exceeded the 28 EU countries combined for the first time in 2016. While these papers have not ranked as highly in the leading publications, the number demonstrates the country's determination in reaching its target of becoming world leading in AI by 2030.
In addition to the start-ups, the established AI players are setting the trends in voice and facial recognition and target advertising, delivering solutions that in many cases surpass those of their US rivals. There has been an increased move within China to use open-source platforms. The Next Generation Artificial Intelligence Development Plan in particular advocated the concept of open-source sharing, and promotes the concept of industry, academia, research, and production units each innovating and in principal pursuing joint innovation and sharing. Such an approach, where companies can share information easily means they will be able to quickly replicate advanced algorithms. In April 2017, Baidu released a code base, which it claims can enable a person to assemble a vehicle capable of (limited) autonomous driving in just three days. This sharing culture is fostering an open AI-knowledge base and speeding up the development of technology in China.
China's strong venture capital ecosystem provides readily available funding for start-ups and entrepreneurs. Venture capital investment increased by 15 per cent. in China in 2017 with investors providing a record $40 billion into over 200 Chinese AI companies.
One start-up, Face++, received the largest AI venture capital fund investment ever for its facial-recognition technology by a C-round investment of $460 million in November 2017.
The big players in China are also leading the investment in AI. In October 2017, Tencent Holdings led a $4 billion funding round in China's biggest on-demand internet service provider Meituan-Dianping. Overall China accounted for five out of ten of the world's largest venture capital investments in 2017, with the US taking the remaining five.
However, they are not stopping there, the big Chinese tech players are also looking to improve their AI offerings by expanding in the US to utilise the US tech talent pool. In 2014, Baidu was the first of the big players' to set up a research centre in Silicon Valley. In 2017, Tencent followed suit and established an AI lab in Seattle while Alibaba announced plans to invest over £15 billion in global research development centres to enhance machine learning, visual computing and natural language processing across seven cities worldwide, including San Mateo, California, and Washington D.C.
While currently in the lead in AI development US regulatory moves towards a privacy by design approach to data and technology is just one reason it could cede that position to China. Similarly, in the EU, while there are some countries, such as the UK, which have the potential to make fast progress in AI development, the focus on privacy, such as the soon to be right 'not to be subject to a decision based solely on automated processing' included in the EU's General Data Protection Regulation (GDPR) is also likely to be seen as restrictive to future AI development .
The US and EU could struggle to match not only the funding and other resources the Chinese government is investing into AI but also the volumes of valuable data being generated.
Has China, therefore, found the right balance to succeed in leading AI globally or will their relaxed approach to regulation cause issues for it in the future?
Written by Rachel Pereira, Trainee, London TMT Group