AI in Banking Systems
Some people think the current banking system will stay stable and unchanged for a relatively long time because of how well-designed and mature it is, but the truth is cutting-edge technologies such as AI should be deployed as a supporting tool to help make banks competitive and relevant for customers, especially in the areas of fraud prevention, risk control, and customer communication.
Firstly, transformation towards AI in the banking system has several benefits:
- Free up human capacity for creative tasks: AI can do time-consuming and repetitive work for human beings — such as scrutinize transactions, customer behavior analysis, fraud analysis, and account management — giving people more freedom and capacity for creation and ideation. The time and energy saved can be used in finding high-quality investment projects, maintaining customer relations, and performing internal and external control, which are obviously more valuable than repetitive tasks.
- Easy to scale: Scalability is arguably an essential part of success, because the larger the dataset is, the more accurate the prediction results will be, and during the fraud detection process the banks also need to process massive amounts of data. Getting AI processes to scale is relatively easier with all the algorithms and programs that are known.
- Easily Adapted: The AI algorithms are highly adaptable, which is convenient for banks to control mistakes arising during the analysis.
- Reduce Costs: To perform customer due diligence and monitor transactions, banks must collect and analyze massive amounts of data, which is costly. For example, the compliance costs of small banks account for an average of 7% of their non-interest expenses, most of which is the salary of employees, followed by data processing, accounting, legal and consulting fees. If banks use AI to cover these processes, the cost can be reduced.
- Better understanding of data & customers: By researching and analyzing the account information and browsing history of customers in the official websites and mobile APPs, banks can uncover exactly what their customers prefer or are interested in, which allow the banks to gain a better prediction power they would never have otherwise.
Besides the advantages mentioned above, AI also has other benefits including streamline processes, improving service quality, and reducing human errors. We believe that its development will bring more and more benefits to cause a profound impact on the banking industry.
However, there are also some potential risks：
- Dependency on third parties: Machine learning uses algorithms to process large amounts of data and recognize the “pattern” within it. Therefore, after a certain point, the algorithms are created by the data. However, banks, the owner of the data, will have to depend on third parties to debug and adjust the algorithm performed on their own banking systems. How to define the owner of the resulting algorithms and can banks have access to the (potentially) proprietary algorithms have become complex questions.
- Regression errors and misled reasoning: When judging whether an activity is suspicious, the bank does not consider gender, race, or marital status, but artificial intelligence is taught by historical data, which may unintentionally imply a violation of the bank’s policies.
- Trade-off: data protection vs. helpful solution: AI programs require access to vast amounts of data to “learn” and work well, and in the banking context, the data set is likely to include sensitive customer information, which may cause a violation of customer privacy.
Considering all the risks mentioned, with the rapid development of AI, it’s clear that potential benefits from implementing AI will out-weight adverse effects.