China's loose data protection rules mean it could catch up and even exceed the US when it comes to artificial intelligence solutions based upon client needs. But it will need to overcome several other obstacles before it is likely to become a genuine international leader in the AI space, say experts.
Currently, China offers a key advantage when it come to AI evolution: its rules are relatively loose. Beijing introduced a cybersecurity law in June 2017 to specifically address the collection, storage, transmission and operation of personal information. The law imposes security and data protection obligations on “network operators”, and restricts data transfers by multinational companies outside China.
But the rules are still open to interpretation. Shi Jingyuan, partner at international law firm Simmons & Simmons, said there was a lack of detail about how the cybersecurity law would be implemented, adding that the data privacy part of the law is “less comprehensive” than equivalents in the European Union or the US.
For now this is an advantage. Internet operators effecitvely face no restrictions on how they use consumer data beyond a requirement to obtain consent, said Daniel Celeghin, head of wealth management strategy for Asia Pacific at consultancy firm Casey Quirk. That’s led online services to collect reams of client data.
“As long as China’s data privacy regime remains anodyne, [China’s] immense advantage in the AI space will remain,” said Celeghin.
Yet while this data collection potential offers a big advantage, some key difficulties could stymie the speed of its AI evolution.
BAML’s Ma noted the country lacks standardised data, making it tougher for automation. Additionally, other experts say China will have to wait a few more years to identify accurate patterns from the big data being collected.
AI seeks to identify patterns from the data. If it just predicts things based on data from the past couple of years only, there’s going to be a bias, said Musheer Ahmed, general manager of the FinTech Association of Hong Kong.
Fintech companies in China have yet to fully utilise the vast amount of data that is available. They have just begun to analyse the available data, added Schulte.
“The first generation [of Chinese AIs] will be clunky,” added Celeghin. “They will get more wrong than right. And the adoption rate won’t be very high. But over time, this will get better and better.”
In addition, China also needs more experienced AI analysts and engineers.
“You can’t just talk about the data. You can do nothing with the data if you don’t have the people and the technology to process it,” Ahmed said.
“The challenge for China [in the AI race] is human capital, and this is the challenge for anybody right now,” agreed Kapron. Humans are needed to analyse and understand AIs over the longer term, but must also understand the industry AIs are used in. For example, any engineering student studying AI could code a chatbot for a bank, but would lack industry understanding to best use the creation, he said.
It could take a newly graduated person five years to perfect a bank chatbot, while an industry veteran may require only one or two years to do so.
In truth, this is a problem for the entire industry. Fewer than 10,000 people in the world have the skills necessary to tackle serious artificial intelligence research, according to Element AI, an independent lab in Montreal in the US.
But the US likely has a slight advantage when it comes to the number of smart and experienced financial whizzes who also know how to code, said Kapron.
External investments trailing
China also sits behind the US in terms of external financial investments, while the latter also enjoys many tech hubs, which gives it a recruiting advantage, according to a report released by McKinsey Global Institute in June 2017.
In 2016, the US absorbed around 66% of external investment into AI, China was a distant second, at 17%. The external investments were estimated at between $5 billion and $8 billion for North America, while Asia only had $1.5 billion to $2.5 billion, according to the report. It defined external investments as venture capital investment in AI-focused companies, private capital investment in AI-related companies, and merger and acquisitions by corporations.
There is also a perception that Chinese universities are not that strong, especially in teaching students about AI. That’s one area where Beijing can probably play a bigger role, said Ma at BAML.
The mentality of Chinese consumers helps China’s AIs gain data too. While many US citizens don’t like to surrender too much personal data, Chinese citizens don’t generally have that “sense of distrust” and surrender their information voluntarily, said Paul Schulte, founder of Hong Kong-based research house Schulte Research.
The country’s middle class in particular use their cellphones for many daily activities. The number of internet users in the country reached a record high of 731 million users in 2016. Of these, 695 million people access the internet through their phones and there are 469 million mobile wallet users in the country, according to data by the China Internet Network Information Centre.
“You are going to be surrendering your location data on your cellphone, other information like your private address, or your spending habits. AI is when computers can look at hundreds of millions of people…and figure out consistencies, patterns, if you are who you say you are,” Schulte said.
But these disadvantages are surmountable. And the ability of fintech companies to access customer data in China offers it far more potential to develop AI solutions than corporations operating under far stricter data protection rules in the US or elsewhere.
By using algorithms to look at data such as individual investors’ financial details, financial advisers or wealth managers can better understand their true risk tolerance, and so create products that are most relevant for them, said Celeghin. And asset managers could develop highly accurate profiles of individual investors from a financial perspective, mapping out future financial needs, and proactively distributing tailored portfolios, he predicted.
For example, a husband and wife who send their six-year-old son to an English-language school might well look to send him overseas for college when he is 18. Knowing this, financial institutions could develop and propose a 12-year lifespan portfolio, recommending how much the parents put away every month, projecting the expected return—and updating them regularly via mobile message.
“I think that’s going to be something that we will see in China over the coming years,” Celeghin said.
But that is some way away. In the shorter term, Beijing needs to collaborate with educators and tech companies to get more brainpower supporting this flourishing industry. Only then can it boast of being an AI superpower.
This is the second article in a two-part focus on China's engagement with artificial intelligence, adapted from a in-depth looking into this topic in AsianInvestor's December 2017/January 2018 magazine. Please read part one here.