Grace Haaf, Assistant Professor and Faculty Fellow of Business Analytics at NYU Shanghai, looks at how the risks and opportunities created by Artificial Intelligence will impact on on business in China
The ever-wider use of Artificial Intelligence is undoubtedly the next step for business development across the globe—it promises to revolutionize the speed at which we can create products and analyze data while also having potentially serious ramifications for employment and economic growth that have specific implications for China.
In this interview, Haaf explores the effects of AI on business in China, focusing on how companies need to adapt in order to be successful in the new AI age.
1.How does China rank in terms of AI development and how does it compare with other major AI developers?
Comparing the conversations that I’ve had in the United States, working at tech startups both in New York and on the West Coast, and the conversations that I’ve had with students and other members of the community in China, I believe China is leading the world in AI development. There is a largely Western perception that Silicon Valley is the cradle of the tech civilization in terms of capabilities and funding, but with regards to technical expertise and enthusiasm for starting companies, the conversations I have in China are in fact very similar to those in the US. I have the feeling that society as a whole is more interested in tech and AI in China than in the US.
2. How and why does China have an AI advantage?
As is often the case in China, there appear to be more resources available, multiples more of well-educated, incredibly talented people available to do AI-focused work. Domestically there are many programs popping up, both at Chinese and joint international universities, and also—for a myriad reasons—students who want to train abroad and then return to China.
Relative to other countries with which I have experience, I have observed more advanced data collection and analysis capabilities in China. Naturally, this plays into certain surveillance stereotypes, but I’m looking at it more from the perspective of a consumer. The integration of China’s finance, social, and technical systems makes the customer experience much smoother compared to what I see anywhere else.
Apps are so integrated with one another that I haven’t touched cash once in the last year. Booking flight tickets, auto-translating taxi driver text conversations from Chinese to English, and even paying for fruit at trail-side stands on a remote mountain hike can be done in a single finance-social online platform, WeChat, that’s connected with my bank account at a traditional banking institution. When you have systems that talk to each other like that, you have the opportunity to build rich data sets in more ways than when you’re just fighting basic IT issues in order to collect information and synchronize it across multiple platforms.
3. Apart from companies like Tencent and ByteDance that seem to be made for an AI world, how are more traditional Chinese companies coming to terms with AI?
The Chinese banks are an excellent illustration of traditional business sectors that have successfully adapted to rapid technological modernization. The finance industry embraced the ability to integrate legacy IT systems across many different third-party platforms, resulting in a near-seamless experience for the customer.
I think the general perception of AI in business is that companies are sitting on omniscient, super-intelligent entities—almost like oracles; however, in reality, most companies are still sorting out the backend IT and internal governance kinks in order to get all their data in one place. Companies are marketing this process as “AI”, but at the moment it’s just a slight improvement in analytical capabilities as a result of having more data to work with plus partial-automation of basic operations like sending someone a coupon based on a previous product purchase. Traditional industries in China, compared to the rest of the world, are moving along this development curve faster because they’re investing in the foundational backend IT and data sharing infrastructure development. It’s not the sexy part of AI, but it has to be done.
4. Are there any models in China that you think would especially be worth emulating in other markets or by other companies?
Digital payment systems and supply chain integration. For the US, COVID-19 accelerated the already in-process development because there was strong demand from the consumer side for contactless payment coupled with regulatory pressure, but it’s still not on the level that China is on.
Anecdotally, living in New York during the shutdown, the most innovation I saw in both digital payments and online/offline data integration came from Starbucks, not from the financial institutions. Within a week of shutdown going into effect, Starbucks first identified a subset of stores that could serve coffee to the sidewalk from the windows or doors without customers entering, second, rolled out search capabilities through the app to find the available locations on a map with 100% accuracy in a rapidly changing situation, and three, seamlessly combined in-app ordering and payment for virtually all customers.
China had food delivery and supply systems already in place when COVID-19 hit such that the government could impose stronger restriction on movements on larger scales without the fear of food and basic necessity shortages. Even in a city of 21 million people, like Shanghai, the government and private sector in combination were able to effectively identify the locations of individuals, design hyper-local distribution systems via neighborhood or building security to minimize contact, and deliver sufficient amounts of supplies, in some cases automatically without individuals placing orders.
5. What is your vision of how access to AI and digitalization will develop and over what time scale, particularly with regard to China?
We’re already seeing massive wealth polarization in societies globally, which has been accelerating since the 2008 financial crisis. Unsurprisingly, we’re also seeing a rise in political populism. It seems that there is a clear divide between the haves and have-nots. You are either born lucky enough to have realistic access to a good education, which in turn would put you into the top jobs, or you aren’t.
As things get more complicated in terms of AI and digitalization, there is less and less access to opportunity for people in low-skilled jobs or with low formal education. As it stands, you need to have had at least four years of education to participate in the new AI workforce. It’s realistically closer to ten years at the top end where you need a PhD to do boundary-pushing research. The more we increase the complexity of the systems we’re dealing with, the more we’re going to drive a wedge between those who were lucky enough to be born into the right situation and those who weren’t.
I think that the major beneficiaries of almost any economic development or innovation are always going to be the wealthiest, proportionally-speaking. I see the talk of Universal Basic Income (UBI) coming from the ultra-rich and I think it’s because they fear revolt against them.
From an AI policy perspective, it could go either way at this point. A country could opt to proactively implement policies that use AI to rebalance society. The US, for example, could end tax breaks to companies for developing warehouse facilities that plan to use mostly automated labor vs traditional low-skill labor. Even if the wealthy still disproportionally benefit from AI development, at least the growing wealth divide is somewhat curtailed.
I think China faces similar economic challenges as the US, but it seems to be doing a little bit better on basic access to resources at scale for the people than the US is doing, so perhaps it will continue that trend with AI-related policy as well. There appears to be more effort aimed at modernizing the workforce in China. Even remote, rural villages have been brought along over the last 10 years with dramatically improved access to mobile data and high-speed internet. I think we’ll continue to see increasing instability and resource polarization in the West. For China, the government might be doing enough to ease the technological transition in a way that is economically manageable for most people.
6. How should companies adapt their business models to prepare for the impact of the AI revolution?
There is always a new tech-marketing word du jour—big data, cloud computing—and now it’s AI. As far as we know, no individual or company is sitting on capabilities so advanced they almost have a robot walking around that can pass for human. That level of tech is just not happening. For most companies, the AI revolution is going to mean improving backend IT infrastructure; integrating data sources and related business processes across various functions like marketing, operations, and finance; and upskilling the workforce from conducting data analysis in Excel to using a programming language.
Don’t panic! Companies don’t need to be able to build self-driving, flying cars tomorrow to be competitive. The good news is that there will be time to react to the introduction of AI, but companies do need to react. Invest in technical skills training and cross-functional communication. This has always been true for developing expertise in specific functions, such as operations or marketing, but as these concepts increase in technical complexity, it is even more important that people can talk across disciplines.
I often see senior management—understandably—get nervous about trusting the analyses of their junior technical experts. It’s hard to make a billion-dollar-bet when you don’t even have the language to ask the questions you’d need for a reasonable risk-reward decision assessment. This is analogous to a brain surgeon trying to understand their true exposure to AIG’s credit default swap positions during a mortgage application process with their local bank. Brain surgeons are smart people who know medicine, but there is no way they’re going to predict the 2008 meltdown from talking with their local mortgage agent. Companies will need to invest most of their time in training senior decision makers how to ask questions and junior technical experts how to respond in a business context.
7. AI can be used for good or for evil. Is there any practical way to ensure that it falls in the right direction?
I ask myself this question all the time. It will have to be collectively driven since you can’t depend on those with elite levels of power and money to not use it for evil. However, that isn’t an AI issue—it’s a general social issue. The people will have to keep up pressure on governments, companies, and other organizations for AI to come down on the side of good and not bad. It’s a dynamic system, and there is no magic-bullet policy that ensures only beneficial outcomes for all people all of the time. The outcome will be decided through billions of complex trade-offs presented in ambiguous contexts, so it’s important to bring as many people as we can into the conversation with sufficient knowledge to advocate for “right”.
8. How is AI impacting the financial and banking sectors already and how is that trend going to develop over the coming years?
I think we’ll see more innovation in the retail and commercial banking sectors as opposed to the securities trading side. The trading side of finance has been using AI in one way or another since the 1990s with the advent of algorithmic traders. The twin hyper-competitive and lucrative aspects of that environment forced rapid technical evolution decades ago when the internet first took off. Those players were early adopters, but now we’re starting to see AI hit the mainstream.
The new greenfield ventures are in retail, for example democratizing retirement account portfolio management. Until now, most middle-class investors have had limited options for financial advisory: you find a financial planner at a community bank or trust your company’s choice of 401k investment fund and just hope they know what they’re doing. Now with digitization and AI, an individual’s access to knowledge and quality isn’t as limited by their location or employer. Caution: there are trade-offs on customer service (a robo-advisor won’t “feel bad” if it loses all your savings on a risky bet) and complexity (clients will have to understand AI in addition to finance to make comprehensive risk assessments).
9. Given that China is a manufacturing powerhouse with millions of people employed in the sector, in what scenarios could there be large numbers of workers displaced by AI? Does the solution lie in upskilling?
The biggest risk is to low-skilled workers. Anything that requires limited training before you do the job will be ripe for being replaced, so what do you do with all those people? Potentially one of the most effective options is government intervention to make higher-level education more broadly accessible. I say governments because they’re likely the only entities that have the power and resources to do it at the required scale. The flip-side of that coin is the arms race in education that we’ve seen in the US over the last 20 years. Now it’s not enough to get a high school degree because you’ll be competing against people with a Bachelor’s degree, who in turn will be competing against people who have a Master’s degree. People are upskilling whether or not the job actually needs it, costing themselves both money and income-earning years in the workforce, due to market-pressure rather than technical necessity.
Interview by Mable-Ann Chang