Various banking systems are required to adhere to common standards for conducting their activities, including customer accounting. By starting with a small dataset and building it up, over time, a financial institution will be able to arrive at a one-size-fits-all way of accounting for customers. The modern data management process in the banking system can be conditionally divided into three components: database management systems, modeling of existing data and ETL processes.
Each step requires specialized tools such as SQL Server, Teradata, and Informatica. Some recent financial market players such as Paxata and Trifacta are transforming data specifically for machine learning tools.
Using off-the-shelf tools
Currently, high-level data processing tools are available to companies such as IBM, Microsoft, Google and Amazon. Banking structures in the United States mainly use tools for analysis – Microstrategy, Qlik, Tableau, for machine learning – Heat, for external interfaces – Google Studio, for internal – Google BigQuery.
Process efficiency testing
When data processing tools begin to show their first results, the financial institution needs to conduct A / B testing to verify accuracy and achieve a high level of confidence. For example, if any information is used to extrapolate trends or to predict customer behavior, the analyst must check each of the variables in the regression analysis and the results against the existing dataset. Such AI services to banks are provided by RapidMiner and Feature Labs.
Working with internal resources
How exactly to start using AI in banking is a tricky question. The use of artificial intelligence in banking sector is not as easy as it might seem at first glance. Despite the tangible benefits, smart systems sometimes make mistakes, and in the financial sector, any mistake can cost money and reputation. For example, Microsoft’s machine learning bot, Tau, posted tweets and chatted, but was disabled due to its inability to recognize offensive content.
Before “releasing” an AI product for direct communication with customers, it is worth learning how to use intelligent services for data analysis, for example, to track user churn or count the number of new customers. One such company, sharpIQ, worked with US banks and used current data to predict the number of SBA (US Small Business Administration) lenders. As a result of such calculations, the number of loan applicants has doubled.
A survey of senior executives in financial institutions showed that over the next three years, the most in-demand banking processes that AI will perform will be: risk assessment, financial analysis and project portfolio management.
What’s next for AI in banking?
It seems clear that AI will continue to expand within banks as leaders become more adept at exploiting these opportunities and the opportunities themselves become even more powerful. However, what comes next will depend on the bank, namely on their operating environment. It’s safe to say that the factors that distinguish the successful AI advocates identified in the survey will be prominent among banks:
- Readiness level. How often do banking leaders review AI performance? Do they have processes for supplementing or canceling questionable results? Do they have plans to significantly improve business processes using AI? These are all indicators of the maturity of processes in AI, and they are all areas where AI leaders in banking already stand out from the rest.
- Emphasis on ethics. Ethics is not a new direction for banks. How is AI different in banking? Recognizing that the bank is accountable for the results of the AI process, just as it is responsible for the actions of its employees. Ethical training for technicians demonstrates an interest in the potentially unethical use of AI.