China's AI plans

Deng Xiaoping, architect of the late 1970s and 80s reform and opening up said ‘keep a cool head and maintain a low profile. Never take the lead – but aim to do something big.’ Something the Xi government doesn’t seem to be listening to, except for the big part. It’s oft repeated that since the Deng-led reforms China has experienced unprecedented, unimpeded growth. However, the friction of the ongoing trade war with the USA is slowing the economy considerably, and prodding China to speed up the development of a modern digital economy – one that stresses national technological self-reliance and ‘smart’ everything. One factor that the government believes will help China achieve its development goals is outlined in a three-step plan published in 2017 which describes how China aims to become a world leader in AI (Artificial Intelligence) by 2030.

The plan

In July 2017 the Chinese State Council published a development plan for AI through to 2030. The plan picks out three developmental milestones on that path. In 2020 it plans to have an AI field on par with western nations; by 2025 China plans to have become the leader in AI technology and applications. AI will be applied in manufacturing, medical care, smart cities, smart farming, and national defence. It is planned that this will produce CNY 400 billion from core industrial industries. At this time AI laws, ethics and security will be implemented. The final goal is that in 2030 the country becomes the world leader in AI innovation. The country will have a complete AI theory and technological systems for brain-like intelligence, autonomous intelligence, human–computer intelligence and group intelligence. Improvement of laws relating to AI and ethics around it will be implemented, although it remains to be seen what those will be.

The first milestone date of 2020 was again emphasised by government officials with the release in November 2018 of the Three-Year Action Plan (2018–2020) for AI development in the country. The Ministry of Industry and Information Technology (MIIT) outlined its plan to accelerate the development of a creative AI industry. The ministry plans to select multiple companies to become standard bearers for the new AI development movement, break the development bottleneck and provide robust competition with domestic and international competitors. There are 17 main areas that the ministry is asking for companies to research. These include connected cars, smart service machines, smart drones, medical diagnostic support systems, video surveillance and identification systems. At time of writing the MIIT was accepting company applications to take part in the selection process; the submission process is planned for completion on the 31st December 2018.

Followers of current events cannot fail to have noticed the fear-tinged hostility the US president Donald Trump and his cabinet is displaying to China. This has led to the first exchanges of a trade war that has seen Chinese telecoms vendors such as ZTE sanctioned and almost crippled. The world’s largest telecommunications vendor Huawei’s links to Chinese government and military has been questioned, leading to some companies and countries boycotting their products. This initial skirmish has led China to increasingly look inwards to shore up its defences against its biggest rival through encouraging and sponsoring the growth of self-sufficiency, particularly of the high-tech components and creative processes that will be essential to any realisation of the country’s 2025 AI goal, let alone the 2030 goal.

China has admitted that it is behind its competitors in the knowledge and processes of AI. The State Council Information Office (SCIO) in July 2017 stated that although China had made good progress in AI in recent years – it is ranked second in number of patents and papers on AI – it lags behind in R&D, AI theory, algorithms, equipment, and chips. Another pressing issue was raised and is one that its competitors can also relate to – a lack of talented, trained professionals. Indeed, Tencent says that China needs a million more AI professionals to meet current demands – demands that will only continue to rise as AI continues its growth. It is difficult to see where these professionals are going to come from in the short term, although the recent introduction of non-curriculum AI courses for secondary school students could see an increasing number of young people entering the field with their own innovations and algorithms. The introduction at secondary school of AI classes could lead to a new group of youth with the skills and knowledge to fill the widening skills gap in the country that has resulted in top grade AI graduates earning potentially 800,000 yuan per year (USD115,400 per year) – a huge sum and an increase of 40% compared to 2017. 

Big big data in the Middle Kingdom

The country has many advantages to support its AI hopes, and these are becoming apparent. For starters China has a much larger potential dataset for pretty much everything (including machine learning algorithm training), with estimates that almost 800 million people use the internet regularly, providing a user base to gather data from three times that of the USA’s.

In addition to the number of users with a wealth of data being produced 24/7 the laws regarding data usage and privacy protection by Chinese companies seem laxer than in other countries, which are on a trend towards greater privacy following public backlash against companies using and collecting data without consent. In China however, the public is largely optimistic about AI and the majority don’t mind government or private entities using their data if it strengthens the economy and the country.

Another project in China driven by AI and big data is the Social Credit System already active in 31 provinces and proposed for full rollout in 2020. In this system, each citizen has a ‘trustworthiness’ profile based on a points system that deducts points for socially or financially undesirable behaviour such as speeding, smoking on a train or not paying fines on time. This score can then be used to help those in authority to make a decision about whether to, for example, grant the applicant a loan or mortgage, give the applicant a job, or allow the traveller on the plane.

People can also gain advantages like a no-deposit apartment or free umbrella rentals by improving their credit score through positive acts such as volunteering. By April 2018 the information collected from governmental departments and other sources collated and crunched by AI led to over 4 million of those blacklisted – classed as untrustworthy – by the system from buying plane and train tickets. All of this data already being collected and categorised lends itself to use by AI systems. In November 2018 President Xi Jinping said, “We need to enhance the combination of AI and social governance and develop AI systems for government services and decision-making.”

The great CCTV system of China

An area that has gained China much press coverage is its facial recognition security camera network. As of December 2017, there was 170 million CCTV cameras around the country and another 400 million are estimated to be in place by 2020. In 2018 the Chinese company Sensetime became the most valuable AI start-up in the world with a valuation of USD 4.5 billion. The company specialises in the computer vision algorithms underpinning face recognition. The company says its technology can be used in vehicles to detect driver fatigue, or in advanced driver assist systems such as lane detection or collision avoidance systems. But the company’s most renowned products by far is its facial recognition technology which claims to be able to rate beauty, detect ethnicity, analyse 1000 real time surveillance video streams and correlate faces to databases while monitoring individuals over different time and place. Its face image investigation system SenseTotem claims to be able to detect people from even blurred images and is available on app and website. It has been used by Guangzhou public security bureau and since 2017 has identified 2000 suspects, captured 800 people and helped solve 100 cases, according to Sensetime’s website. Its entertainment kiosk uses facial recognition and feature analysis to provide targeted advertising to kiosk users. The strength and formidableness of the Chinese system came home to western audiences when a BBC journalist tested the system and was caught by the authorities within seven minutes.

Further proof of the prowess of Chinese facial recognition technology was seen in November 2018 when five Chinese companies topped the Face Recognition Vendor Test (FRVT) held by the US Department of Commerce’s National Institute of Standards and Technology (NIST). Top placed company YITU said, ‘the machine is able to recognize every individual almost error-free on a scale of ten million samples.’

That company also provides facial recognition technology to public service providers in China. Examples of criminal cases cracked involve identifying cars running red lights, catching unlicensed drivers in Shanghai, an on-the-run bank manager who had embezzled 1.5 million yuan. Another case solved was that of a man who killed three people before fleeing to another area and eventually becoming Abbot of Longxing temple 13 year later; he was recognized through image matching technology provided by the company. The company’s facial recognition technology has also been deployed on over 12,000 ATMs nationwide; users don’t need a bank card but can use facial recognition to enter their account.

BAT

To develop AI is going to need money. In the West big businesses such as Google, Microsoft and Amazon are investing heavily in AI. And China’s economic scene is dominated by three huge companies collectively known as BAT – Baidu, Alibaba, and Tencent. Each of these companies is actively investing in AI with support and guidance from the government as to which areas each should focus on:

Tencent, makers of messaging platform Wechat, payment service Wechat Pay and owners of many of China’s largest mobile computer games, recently announced that its AI focus was on the development of medical technology for CT scanning and screening for cancers, and pre-diagnosis of 700 types of disease with a claimed 98% accuracy rate. It also focuses on the use for games, multimedia content and human–computer interactions.

Search giant Baidu’s AI research is largely focused on the more fundamental commercial aspects such as facial and body recognition, video analysis, natural language processing, and big data analysis. It must be remembered that Baidu is also a major player in China’s autonomous car environment, and is using its AI research to inform its driverless systems.

Alibaba is working on its own AI chips called AliNPU which is expected to be released in 2019 – hoping to add further self-reliance for chips. Alibaba has also used its AI system City Brain to reduce emergency vehicle response times by 50% and automate traffic signal control, increasing average travel speed by 15%. It also has several more Alibaba Cloud based Brains, including Medical Brain to assist hospitals, doctors and nurses to provide better care; Industrial Brain; and Environment Brain to monitor industrial waste and environmental pollution. Alibaba also invests heavily in other AI companies including Sensetime.

Problems

With the government’s use of AI to strengthen its own powers of monitoring and control through facial recognition systems, big data analysis and the Social Credit system it would be difficult to see what could impede China’s progress to AI dominance by 2030. The systems brought about by AI reinforces government control and authority, making it harder for those that might wish to overturn it to do so. This means government overthrow or instability is increasingly unlikely in the near future.

One possible source of concern for Beijing lies in its increasing inability to remain a watcher in the wings. As China continues its rise, albeit more slowly than previously, it will continue to face international political, military and trade issues. This is exemplified by the current trade war with America and the consequent distrust of Chinese-government-linked entities by other Western countries, because of the perceived risk of spying and intellectual property theft.

In November 2018 the Trump presidency declared that there would be limits on foreign investment in US AI companies, among others, in reaction to perceived imbalances between Chinese opportunities in the US market and those of the US companies in the Chinese market.

One potential indicator of the early effects of the dispute can be seen from the fact that Chinese industrial profits shrank from 21.9% growth in April 2018 to just 3.6% as of November 2018. With a slowing economy the government will have less money to invest in its AI push, making it potentially slower than it would be otherwise. This could be a potentially large issue as the government legitimises itself to its people through its economic growth statistics.

If as some believe the trade disputes will seriously affect China and its economic progress, then perhaps there won’t be the resources for it to grow to AI global dominance by 2030. But if China manages its growing stature on the international stage well and makes friends without alienating too many, then it would be a good bet that China can achieve significant progress towards its 2030 goal.

Add this: