Fintech competitiveness depends on AI technology
By Lee Jung-woo
The explosive growth of the Chinese fintech industry, represented by Alipay, almost causes fear in Korea and all over the world. The growth has come from the immaturity of the traditional financial industry and its giant market, but the easing of regulations by the Chinese government also plays a big role.
In Korea, despite the government’s efforts to ease fintech regulations, many remain, dragging the whole industry down to lag behind China.
Limiting individual investment in peer-to-peer (P2P) financing at 10 million won ($9,220) is a typical one. The ban on non face-to-face contracts on discretionary investments in the asset management field also limits the domain of fintech startups online. It is necessary to change perspectives in modifying regulations to something that will help new fintech companies.
Fintech can be classified into three areas: well-known money transfer and payment; P2P finance represented by cloud funding; and asset management represented by robo-advisors. The common technology necessary for all three is artificial intelligence (AI). Fortunately, thanks to the kick-start policy of the government, colleges and companies are actively executing research and development in AI.
There are three methodologies in AI: supervised learning, unsupervised learning and reinforcement learning. Supervised learning is a technology that gives questions and answers at the same time in the learning process, and unsupervised learning is a method to teach by providing only questions without answers.
Supervised learning is already in use in various fields, such as facial recognition, voice recognition and language translation.
On the other hand, unsupervised learning still remains as underdeveloped technology, with as yet no commercially successful case, but it has a huge potential for progress. Since it is similar to a process of a child learning something, AI research and development in the future will likely concentrate in this field.
Reinforcement learning is widely known, as it was used in Alpha Go, a Google developed AI program. It uses rewards, instead of answers, to make the machine make decisions sequentially and learn by making multiple mistakes. This technology is currently showing progress in games and robot driving.
All three methodologies can be used in three areas of fintech. In transfer and payment, fraud detection system can be powered by supervised or unsupervised learning to detect abnormal transactions.
In P2P lending, supervised learning can be used in P2P for credit scoring and anticipation of expected returns. For asset management firms, reinforcement learning can be used for automated portfolio building.
Unfortunately, there are not many companies in Korea actively using AI technologies. This is because the pool of AI experts is concentrated in sending talent to fields other than fintech, while the pool itself is very small. This causes lack of AI technology experts who can apply their expertise to fintech industry here.
For the growth of the Korean fintech startup ecosystem, the government should lead research and investment in AI. Since driving domestic fintech startups to advance into overseas markets is significant for the growth of those firms, the government should play a more active role by helping companies introduce AI technologies aggressively.
For example, to process big data, a giant graphics processing unit (GPU) cluster is required. For startups, which lack the necessary capital, a cloud computing service designed for AI learning at affordable costs would be helpful.
In order to narrow the gap between Korea’s fintech industry and that of other advanced countries, the government’s integrated and aggressive support for research and development of AI-fintech application technologies, human resource nurturing and infrastructure improvements are required desperately.
Lee Jung-woo is a professor at Seoul National University’s Electrical and Computer Engineering Department. The Fintech Center Korea contributed to this article.