Data science has been at the core of Ant Financial since its beginning. As the company has evolved form beyond a payment platform to a multi-faceted financial solutions platform, they have begun to leverage the data they collect with machine learning techniques in order to develop new products and services for their customers.  Ant has access to an incredible amount of data generated through their core business: payments. Not only does the company have access to financial data, but also to their preferences, through Ant’s relationship with its former parent company, Alibaba. By combining these two sources of data Ant is able to assess the credit of their consumers, and then use that data to offer additional products and services connected to the broader Alibaba network.  AI and machine learning are crucial to generating that incremental additional revenue from each customer, as they allow a company to target resources, advertising, computing power, etc, to a user who is most likely to use that service. Without this ability to optimize, Ant and Alibaba risk being left behind by their competitors, primarily Amazon in the US, and Tencent (WeChat) in China, their home market.
Ant’s use of machine learning goes far beyond simply optimizing their product offering. One interesting example is how they are leveraging their research into computer vision to help automate insurance claims (upload a picture of the damage to be assessed).  One of the other key short-term innovations developed by Ant is the Yu’e Bao money market fund, which has grown into the largest in the world. Ant noticed that there were significant cash balances in user accounts and developed a product that could offer more attractive interest rates to those users due to the availability of data.  However, one of the more groundbreaking solutions that Ant has developed is their credit scoring system, Sesame Credit. Because China lacks a traditional credit scoring system, individuals are often unable to access credit, and will have to put down some sort of deposit etc in order to say, reserve a hotel room.  Sesame has developed a proprietary scoring mechanism that infers a customer’s credit from their previous purchase histories and levels of spending.  Though there are substantial financial incentives to keep a high score, ranging from expedited loan applications to having access to loaner umbrella’s at libraries, customers with low scores are severely punished, with individuals even banned from air travel for defaulting on loans and then subsequently having a low score.  As the use of these types of mechanisms become widespread, Ant opens itself up to additional risks. Compared to a credit scoring agency in the US, where the methodology is reasonably well stated, a machine learning algorithm is less-straightforward to understand, leading people to potentially question the motives of the company.
One of the crucial aspects about Ant financial that separate it from other companies using machine learning is the simple fact that it is a financial company. The ramifications of getting things wrong at Ant are more severe than at Alibaba; an unfairly low credit score can bar people from owning a home or even traveling, a bad product recommendation is not as consequential. So, my recommendation would be to work very closely with regulators as these technologies are being developed. One example of Ant quickly responding to regulator concerns was with respect to Yu’e Bao. Regulators were concerned that the product was growing too quickly, and that there were not enough safeguards in place for customers. Ant has responded by introducing new products, to lessen the size of any individual fund, as well as to introduce dynamically managed daily total subscription cap.  This responsiveness has helped Ant be able to shape some of the necessary regulations that are being introduced.
Some questions I would have for my classmates would evaluate the morality and unintended consequences of innovations such as the social credit scoring at Sesame Credit. The data will help allocate resources and credit to the most productive and credit worthy individuals—society will be more productive as a result. But there will be individuals left behind due to their low scores. Is this an acceptable price for society to pay? What are the ramifications of those products, and are the outcomes that are generated something that we actually want as a society?