Summary of Key Points
This news article focuses on the application of AI in the banking industry and the transformation of retail services. The main message is that AI in banking has moved from being a topic of hype to practical implementation, with the focus of competition shifting towards the depth and effectiveness of specific use cases. Retail services serve as a stabilizer for banks during economic cycles, but they are currently facing challenges in terms of customer structure, business models, and traffic sources. Additionally, the industry needs to overcome three misconceptions: that AI will replace employees, that AI investments will generate quick returns, and that general-purpose large models are omnipotent. The key to AI transformation lies in organizational adaptation and the tailored use of AI for different scenarios, ultimately leading banks to shift from extensive expansion to competition based on specialized capabilities.
I. Banking AI: From “Having Large Models” to “Whether Scenarios Can Be Successfully Implemented”
In the past, banks might have discussed AI as a slogan or shown off concepts, but now we are in a phase where it is actually being put into use. Liu Jianjun emphasizes that the competition is no longer about whether a bank has large models, but whether those models can be applied to specific scenarios such as intelligent customer service (answering common customer questions), marketing recommendations (proposing suitable financial products), and risk warnings (detecting loan defaults in advance). These scenarios share one common characteristic: they do not involve major risk decisions. Therefore, the potential for AI to provide incorrect information (or “delusions”) is low, making them easy to implement and capable of directly helping banks save manpower and improve efficiency.
II. Retail Services as a Bank’s “Ballast Stone,” but Undergoing Triple Challenges
Retail services are crucial because they help stabilize bank operations regardless of the economic situation. Why are they so important?
- Less Capital Required: Retail activities (such as personal mortgages and consumer loans) require 70% less capital than corporate lending. With the same amount of money, retail can generate three times more business.
- More Stable Profits: Non-interest revenues from fees and financial products reduce a bank’s reliance on the interest margin between deposits and loans, which is becoming narrower due to declining interest rates.
- More Stable Funds: The deposits from a large number of individual customers are low-cost and stable, reducing the risk of customer churn.
However, retail services are currently facing three challenges:
1. Difficulty in Segmenting Customers: High-net-worth clients require professional financial management, while ordinary consumers need affordable basic services (such as payments and small loans). Banks must meet both needs, which is demanding.
2. Need for a Change in Business Models: Previously, banks focused on selling products (e.g., credit cards); now, they need to help customers manage their risks over the long term, requiring higher levels of expertise.
3. Loss of Traffic Sources: In the past, customers used bank apps; now, they may obtain financial services directly through AI assistants (like ChatGPT) or other intelligent platforms, eroding banks’ advantages in traffic acquisition.
III. Three Misconceptions about AI Transformation
Many banks have misunderstandings about AI transformation. Liu Jianjun highlights three common misconceptions:
1. Misconception 1: AI Will Replace Employees on a Large Scale: AI mainly replaces repetitive tasks (such as data entry and organization), not high-skilled roles like customer managers, whose core responsibilities involve building trust and emotional connections with customers, which AI cannot replicate.
2. Misconception 2: Quick Returns from AI Investments: Implementing AI is not just about buying software; it requires a long-term effort. Poor AI outcomes in many banks are often due to outdated organizational structures, lack of employee training, unoptimized business processes, and unsuitable operational models. Expecting returns within a year is almost impossible.
3. Misconception 3: General-Purpose Large Models Are Omnipotent: While large models like ChatGPT are powerful, banks need “vertical small models” trained with their own financial data to solve specific problems (e.g., credit risk assessment). These models should be used differently for low-risk tasks (e.g., automated customer service), medium-risk tasks (e.g., AI-assisted small loan approvals), and high-risk tasks (e.g., manual decision-making for large corporate loans).
IV. Banks’ Business Models Need to Evolve: From “Expansion” to “Specialized Competence”
In the past, banks made money by growing in scale—more assets, more customers, and higher revenue. This no longer works due to declining interest rates, tight capital markets, changes in customer behavior (Generation Z prefers digital channels), and shifts in how customers acquire services. AI technology has matured, making it possible to automate processes and enhance collaboration between humans and machines. These developments represent a fundamental shift in banking business models. The future competition will not be about size but about specialized capabilities, such as the ability to serve different customer groups accurately, improve efficiency with AI, and maintain customer trust.
Liu Jianjun emphasizes that the cost of AI transformation lies not in purchasing technology but in adjusting organizational structures (e.g., revising processes and training employees). Moreover, compliance and risk management must remain a priority at all times.
Conclusion
The banking industry is at a critical juncture where AI is moving from theory to practice. Retail services play a stabilizing role, but they need to adapt to new challenges. Banks must avoid common misconceptions about AI transformation and shift from an extensive expansion strategy to one focused on specialized capabilities. The ultimate goal is to maintain customer trust, deepen the use of AI in specific scenarios, and rely on professional skills to navigate economic cycles.