From AI Chatbots to AI Bank Clerks... Data Acquisition Crucial to Enhance Utilization (Comprehensive)

Banks Advancing Use of AI Chatbots (Including Virtual Bankers)
"Enhancing AI Utilization Requires Expansion of Training Data and Experts"

[Asia Economy Reporter Park Sun-mi] Customer: Please tell me how to lift the transfer limit on a new account.

Chatbot: What do you mean? An account with transfer limits?

Customer: Yes. An account with transfer limits.

Chatbot: If you have not submitted supporting documents but the purpose of opening the account is recognized, you can use it as a financial transaction limit-restricted account. I will tell you how to lift the financial transaction limit restriction.

Customer: What documents are required to verify the purpose of the financial transaction?

Chatbot: Please refer to the example documents below and prepare appropriate supporting documents for the financial transaction.

As domestic banks focus on strengthening their digital competitiveness, they are advancing the use of artificial intelligence (AI) chatbots (including virtual bank tellers). AI chatbots, which have been actively used in mobile banking, will be fully deployed in bank branches next year in the form of virtual bank tellers to handle simple tasks previously done by humans.

According to the banking sector on the 15th, Shinhan Bank plans to more than double the number of AI bank tellers to about 200 by early next year. This will cover 73 branches equipped with digital desks, including 4 Digilog branches, 10 digital lounges, and 1 convenience store innovation branch.

Currently, AI bank tellers provide simple guidance services such as checking customer queue numbers and verifying IDs at digital desks. Starting early next year, they plan to upgrade the AI bank tellers’ capabilities to provide basic financial services such as simple inquiries and transfers.

KB Kookmin Bank opened an AI experience zone in the Yeouido new building in March this year and is testing the AI bank teller’s tasks. An AI avatar resembling a person on the screen communicates with customers and handles basic banking tasks such as providing information on pre-programmed financial products. Although Kookmin Bank has not yet deployed AI bank tellers in branches, it plans to enhance the system next year and pilot it in some branches, gradually increasing the number of deployments.

KB Kookmin Bank AI Experience Center

NH Nonghyup Bank, which officially introduced AI bank tellers last month, has deployed them in 13 branches in Seoul as of the 3rd. Nonghyup Bank plans to assign AI bank tellers personnel numbers and issue official appointments aligned with new employee hiring schedules, treating them like regular employees. From January next year, they will undergo about three months of training and probation alongside the 2022 batch of new hires.

A bank official explained, "In the short term, AI bank tellers will assist mainly with one-way explanatory guidance, but in the mid to long term, the goal is to provide services capable of two-way communication. They will develop to a level that can replace a significant portion of the tasks performed by human bank tellers."

Before deploying AI bank tellers in branches, the banking sector first introduced AI chatbot consultation services in mobile banking, receiving positive feedback on their effectiveness and usability in the non-face-to-face era. Currently, AI chatbot consultations in banks can handle simple banking services such as transfers and inquiries, recommend financial products, and provide guidance on required documents.

The banking sector ranks chatbot investment as the top priority for AI utilization. According to a survey conducted by the Korea Institute of Finance from May to August this year targeting eight domestic banks, chatbots were the most common area for planned additional investment, followed by credit evaluation, loan screening, financial asset recommendation and management, and digital marketing. However, challenges faced in AI adoption include lack of data (25%), shortage of related experts (21%), and insufficient short- and long-term implementation strategies (13%).

Seo Jeong-ho, a research fellow at the Korea Institute of Finance, emphasized the need to secure training data for AI system development, stating, "AI personnel identified 'lack of training data' as the biggest challenge in AI development. Although AI is based on big data, bank-internal data often have low transaction frequency or are unrefined, making them unsuitable for algorithm development. For external data, the data integration process is complex and time-consuming, which is also a challenge."

He added, "Recently, banks have been actively promoting AI adoption in various fields including chatbots. For AI utilization to lead to productivity improvements in domestic banks in the future, it is necessary to establish internal management systems related to AI development and utilization, expand training data and expert personnel, and formulate and execute consistent AI strategies."

This content was produced with the assistance of AI translation services.