[Interview] "AI Should Focus on Quick Problem Solving, Not Replacing Counselors" [Bbaengbbaengi AI Call Center] ⑩

Interview with Im Euntaek, Head of AI Unit, Digital Innovation Division, Shinhan Bank

Editor's Note“It’s a ‘person,’ not AI, dear customer. How can I assist you?”
Call centers, which everyone has used at least once, have long been places where agents kindly answered calls and resolved issues. However, financial company call centers have gradually become inconvenient for financial consumers. The incomplete ‘Artificial Intelligence (AI) consultation service’ has emerged, but it has instead delayed the problem-solving process and increased the time consumers experience. This is the irony that has arisen as banks, card companies, and other financial institutions replace human agents in call centers with AI consultation services. Accordingly, Asia Economy aims to look into the inside story of financial company call centers from various perspectives, including those of financial consumers and workers.

“One of the mistakes companies make regarding AI consultation is the belief that AI consultation should replace human agents. The key is how quickly AI can help solve customers’ problems. This is the customer immersion that Shinhan Financial Group pursues.”

Lim Eun-taek, Head of the AI Unit at Shinhan Bank’s Digital Innovation Division, said in an interview with Asia Economy on the 18th of last month at the Samsung Headquarters Building office in Jung-gu, Seoul, “Shinhan Bank’s AI consultation service focuses more on how customers can quickly resolve complaints rather than AI itself.”

Im Eun-taek, Head of the AI Unit at Shinhan Bank Digital Innovation Group, is being interviewed at the Shinhan Bank office in the Samsung Main Building, Jung-gu, Seoul. Photo by Jin-hyung Kang aymsdream@

Regarding concerns that the full-scale introduction of AI consultation services is reducing call center agents and that this could lead to consumer inconvenience, Lim explained that it would rather be an opportunity for the emergence of ‘specialized consultation personnel.’

He said, “From a business perspective, if AI replaces the work of 100 people, the number of agents will decrease accordingly. However, from the customer’s perspective, there is a downside that waiting times may increase due to the reduction of human agents when customers want to have difficult consultations rather than simple ones. In the future, rather than viewing AI consultation services as reducing people, I think the roles will change so that AI handles consultations well and humans take charge of ‘specialized consultations.’”

Shinhan Bank is classified as one of the most proactive banks in the financial sector regarding digital transformation, including AI. In a sentiment analysis of AI consultation services conducted by Asia Economy through Data & Research on the five major commercial banks, Shinhan Bank was the only bank showing an upward trend in net positive sentiment.

Accordingly, the proportion of AI consultations in Shinhan Bank’s call center is steadily increasing. For outbound calls (calls made proactively to existing or potential customers), about 98% are handled by AI. Of these, about 30%, roughly one-third, complete the task without connecting to a human agent. AI chatbots used through applications also handle about 2.5 million complaints per month, accounting for about 40% of all complaints.

AI Consultation Still Auxiliary... Considering Solutions to Customer Inconveniences Such as Delayed Consultation Times

However, for inbound calls (calls made by existing or potential customers to the call center), the proportion absorbed by AI is about 60%. Among these, the proportion of consultations completed solely by AI is less than half, about 25%. In other words, about 75% of consultation calls to the call center still require human agents.

He explained, “Currently, the advantage of AI consultation is that simple tasks like basic inquiries can be completed very quickly, but the disadvantage is that it is difficult to perfectly handle complex consultations that require various information. For complex consultations or areas where regulations require human agents to obtain consent, it is still easier for humans to handle them.”

Lim is well aware that many financial consumers feel inconvenienced when using call centers after the introduction of AI consultation services. He said, “AI consultation is only an auxiliary means. Complaints are more about the overall reduction in human agents after AI introduction, leading to increased waiting times and difficulties in use, rather than dissatisfaction with AI consultation itself.”

To reduce these customer inconveniences, Shinhan Bank is preparing improvements to AI consultation services on two fronts: providing consultation services tailored to each customer’s emotions and tendencies through sentiment analysis, and advancing the AI consultation system by introducing generative AI, such as Chat GPT, to offer personalized consultations.

Lim said, “Some customers feel inconvenienced because the time to solve problems with AI consultation is longer, while others complain it is too fast. Through the Artificial Intelligence Contact Center (AICC), we plan to add sentiment analysis functions to AI consultation services, categorizing customer emotions into three levels?whether they are angry, neutral, or otherwise?and have AI respond accordingly.”

Im Eun-taek, Head of the AI Unit at Shinhan Bank Digital Innovation Group, is being interviewed at the Shinhan Bank office in the Samsung Main Building, Jung-gu, Seoul. Photo by Kang Jin-hyung aymsdream@

He also said, “Currently, AI consultation services operate based on scenarios and are not optimized for users, but we are researching ways to advance them by introducing generative AI like Chat GPT. We plan to develop AI consultation services so that they can provide more personalized business consultations to customers.”

On the other hand, measures are also being prepared for digitally vulnerable groups such as the disabled and the elderly. Lim said, “When kiosks were introduced in fast-food restaurants, it was true that the elderly had difficulties using them. For elderly customers who find AI consultation difficult, we have established and operate an educational institution called ‘Hakijae (學而齋)’ in Incheon to provide financial consumer education.”

AI Consultation Advancement Will Take About Five Years... Reduction in Agents? The Nature of Human Work Will Change

However, there is still a long way to go. While Shinhan Bank’s AI consultation service covers 100% of simple tasks, it shows limitations in handling complex tasks. Lim expects that it will take several more years for AI consultation services to advance beyond simple consultations.

Lim said, “For AI to provide natural consultation services, a technological level corresponding to artificial general intelligence (AGI) is required. While companies like OpenAI say AGI can be realized within two years, realistically, I think it will take about five more years.”

There are also remaining issues such as social norms and legal regulations. For example, generative AI like Chat GPT lacks established norms regarding responsibility when errors occur, and since connection to external networks is inevitable, network separation regulations remain unresolved.

Lim pointed out, “For instance, if a customer inquires about a loan product and generative AI services like Chat GPT or Naver Cloud provide information based on outdated policies before updates, causing the customer to lose about 1% in interest costs annually, it is unclear who is responsible. Social norms regarding responsibility for errors caused by generative AI need to be established first.”

He added, “There is also the issue of network separation regulations to prevent internal data from leaking externally. Authorities recognize the need to improve regulations and are likely to allow certain levels through innovative financial services, so we are currently waiting for their decision.”