While widely regarded as a finance company, China's Ant Financial is essentially an artificial intelligence company trying to bring more equality to the world, writes its chief data scientist and head of AI, Dr Alan Qi.

Dr Alan Qi

Delivering financial services is all about data, from insurance, payments and investments to risk management and microloans. But the deluge of heterogeneous data from multiple sources makes it challenging to generate business insights, improve user experience, calculate risk or identify new product opportunities, as well as ensure security and privacy protection.

And with hundreds of millions of transactions per day, it is not feasible to use manual methods to make necessary decisions quickly. For example, in the case of real-time payments, decisions must be made in milliseconds to protect against fraud.

Artificial intelligence (AI) is helping to address these challenges. At Ant Financial, we have brought together machine-learning methods – such as deep learning, reinforcement learning and computation graphs – on one platform to support our internal business applications. We are also opening our platform up to external partners and clients in our ecosystem to develop new products and improve customer service. The goal is to reduce cost, increase efficiency and better manage risk, while ensuring the system is secure and works in real time.

AI applications in real life

We are changing people’s lives by applying AI to real-life problems, in line with our motto to “bring the world equal opportunities”. Instead of making the world more polarised, we want to make it more equal to help the majority of people.

For example, we can provide an immediate microloan for small and medium-sized enterprises (SMEs) in remote regions in China. The turning point for the microloan business was the use of big data and machine-learning algorithms. We describe the process in three numbers: three, one and zero. It takes clients less than three minutes to fill out the microloan application form; one second to receive the approval or rejection decision; and it is all done with zero human intervention.

We work with the China Foundation for Poverty Alleviation (CFPA) Microfinance, whose aim is to reduce poverty in rural areas of China by providing loans to farmers. Ant Financial provides the technology and is working with CFPA Microfinance to improve its efficiency and capacity in terms of risk management. Recently, CFPA Microfinance overhauled its microloan process, moving from manual to AI-based assessments. The CEO posted online that the efficiency was improved by 10,000%, which is the essence of what AI technology can bring to the world.

Fraud prevention is another area where AI is making an impact. With hundreds of millions Alipay users online, we need to ensure these are authentic accounts and remove the fraudulent ones. But we have only very little information to make a judgement when users first sign up. Neural networks can provide insights with even a small amount of data and thus we can use machine learning to detect suspicious accounts. With this technology, we have seen a vast improvement in terms of accuracy over the simple rule set previously used to identify spam accounts.

We can also use AI to create new products tailored to specific client segments. For example, we designed an insurance product for young women who purchase skinny jeans. By analysing data, we discovered that women who wear skinny jeans are more likely to need insurance for damaged mobile phones, so we developed damage insurance for mobiles and targeted these users. Anecdotally, a speaker came up to me following a speech at a conference in Beijing and pulled out two mobile phones – both screens were damaged and she was wearing skinny jeans.

Small, simple – and beneficial

This example illustrates how it is possible to come up with many simple, small financial products that might be beneficial to many people.

In another insurance example, a customer can send us video footage showing car damage following an accident and, using AI, we can provide a cost estimate to fix it and also suggest where it can be fixed. Whereas this is a boring task for humans and takes several years’ experience to recognise which car part is compromised as well as estimate the cost to fix it, a computer can make a quick and accurate estimate.

AI is the engine for financial services to make the customer experience better. For example, we have a project related to an intelligent assistant, inspired by a specific customer service problem we faced. We have almost 700 million users in China and it is a difficult task to serve the sheer user numbers with human assistants.

This problem was particularly acute on Singles Day (November 11), the largest shopping festival across the world. Customers are hardly able to reach the customer service people by phone due to the high number of calls. In 2015 we rolled out an AI chatbot to answer questions. Impressively, 94% of problems raised by customers were solved by our intelligent customer service system. In 2016, the self-service rate was 97%. We reached a milestone in 2017 when customer satisfaction with chatbots was higher than with humans.

Ethical considerations

As illustrated, Ant Financial is focused on how to formulate real-world problems into AI solutions. We also use the technology to improve our current systems and the user experience, at the same time cultivating new business opportunities.

A few things happened simultaneously in the past few years to increase the power of AI. First is the developments in cloud technology, graphics processing units and field-programmable gate arrays that have dramatically amplified computing power, which lets us rapidly process a massive amount of data.

The second driver in the rise of AI is the advance in machine-learning methods, which has allowed us to do things that were previously impossible. As such, we are witnessing the next phase of development of AI, both in terms of the technology’s maturity and the enormous impact it will have on business and society.

The debate the world is having around ethics is very important, as we move into this new AI-enabled world. But if we think of AI as the engine in a car, the driving wheel is in human hands. For example, when a machine-learning algorithm is programmed by a person to optimise a goal, the objective function is set by people, not machines.

Looking at it from Ant Financial’s mission to bring equal opportunities to the world, our aim is to use different technologies to solve complex problems in our daily lives. The technologies are the engines, but we decide how we are going to apply them, for what problems and for which users. For example, we use AI to help SMEs improve user experiences for disabled people or the aged – such as using voice capabilities with a mobile phone to order a taxi service, so that an older person does not have to search on a tiny mobile screen.

So we have developed our own guiding principles for using AI – but it is not one line. The Chinese way is to have many lines, tailoring the principles to each business line. And this is an ongoing effort because as we develop new applications, we are also developing new rules simultaneously.

But, at the same time, we believe technologies used in the financial domain should be regulated. Technology is a double-edged sword and integrating it within a regulatory framework ensures a cautious and well-thought-out approach. It is important to be cognisant of the risk, even when trying to maximise the utility for a specific purpose. Like when designing a new car, we all think about how to make it safer, not solely about making the user experience better.

We are also working on explainable AI and developing systems for this purpose. Interpretable machine-learning methods are crucially important for regulatory oversight and risk management, as well as for the fundamental robustness of the financial system.

The future is data

Data science will be a transformative force across society. But it is important to start from real-world problems, rather than a being hammer finding something to crack. Our first question should be: what kind of financials services do our customers need? That gives us guidance in developing technology.

Collaboration is also important and we must work together with partners, regulators, big players and newcomers – domestically and globally – for the good of society. There is much to be gained. I believe the world needs greater communication, rather than separation and protectionism.

Ant Financial invests in technologies we believe can help solve the complex problems humanity is facing. It is important to continuously keep the problems we are trying to solve and the direction of travel at the forefront of our minds, because it is easy to get side-tracked. We want to bring the world more opportunities and therefore see the importance of guiding the technologies in that direction.

I do not agree with the scaremongering around robots taking our jobs. It is true that AI may replace some boring and challenging jobs, such as assessing a damaged vehicle or in customer service. But in the long run, machines will set people free from many repetitive jobs and let them do more interesting and creative work – such as writing books, or building the robots of the future.

Dr Alan Qi is the chief data scientist and head of AI at Ant Financial.

PLEASE ENTER YOUR DETAILS TO WATCH THIS VIDEO

All fields are mandatory

The Banker is a service from the Financial Times. The Financial Times Ltd takes your privacy seriously.

Choose how you want us to contact you.

Invites and Offers from The Banker

Receive exclusive personalised event invitations, carefully curated offers and promotions from The Banker



For more information about how we use your data, please refer to our privacy and cookie policies.

Terms and conditions

Join our community

The Banker on Twitter