AI making a mark in banking & finance

Diverse artificial intelligence (AI) applications have already made a mark on the banking and finance sector, according to RMIT Lecturer Dr. Vo Thi Hong Diem.

Intelligent chatbots and virtual assistants are at the forefront, equipped to understand and resolve customer queries, provide tailored financial services, automate tasks, identify fraudulent activities, assess creditworthiness, and deliver automated customer support solutions.

AI’s integration into the finance sector has witnessed remarkable growth. In McKinsey’s 2019 Global AI Survey, nearly 60 per cent of financial services sector respondents indicated that they had already embedded at least one AI technology to perform structured operational tasks or detect cybersecurity risks.

Meanwhile, a survey by the World Economic Forum in 2020 revealed that 85 per cent of financial organizations were incorporating AI technologies into their operations at the time, while 77 per cent of senior executives anticipated AI to hold high or very high business importance in the subsequent two years.

Following the global trend, prominent banks in Vietnam have invested in researching and implementing AI technologies in their operations.

For instance, TPBank has integrated face recognition technology into its LiveBank automatic banking channel, bolstering security and convenience for customers. VietinBank utilizes kiosks with FaceID recognition to identify customers and forward their requests to advisors, as well as serve as valuable assistants.

Other banks like VietABank, Nam A Bank, VPBank, Techcombank, VIB, and ACB have embraced AI across various functions, including chatbots for customer support and engagement, asset management, security, fraud prevention, and analysis of peak season ATM withdrawals.

The incorporation of AI technology in the banking sector not only optimizes operational costs but also enhances customer support and enables efficient process automation. AI has proven highly advantageous for revolutionizing data management, customer behavior understanding, and fostering robust customer relationships.

However, it is essential to note that most banks in Vietnam employ traditional rule-based AI, which excels in handling routine enquiries and assisting with simple financial transactions. This type of AI can only automate tasks that have been programmed into it, and its training is usually tailored for specific stationary tasks, making it less adaptable to new situations or tasks.

In contrast, generative AI possesses the ability to be trained in a wide range of data and can adapt to various situations and changes, but its application in the banking sector remains limited.

Generative AI stands as a next-generation technology that takes automation to a higher level by empowering computers to generate fresh content and ideas, moving beyond mere data processing and analysis.

The significant difference between traditional AI and generative AI is their learning and adaptive capabilities. Generative AI can process past data, learn from it, and make intelligent decisions based on this knowledge, while traditional AI is confined to performing tasks designed for it.

Generative AI can continually re-train, update, and adjust its predictions, diagnoses, and decisions in response to new data inputs. This adaptability aligns with the increasing demand for personalized financial services driven by customer preferences.

Furthermore, generative AI can access the information necessary to undertake complex tasks related to customer information, and complete simple or complex automated payments as an autonomous AI agent without human supervision.

The integration of generative AI into the banking sector in Vietnam currently faces several challenges that impede its widespread implementation.

Firstly, Vietnam lacks a solid AI development ecosystem and appropriate support policies, placing it at an early stage compared to some other Asian countries.

The high cost of AI and advanced machine learning and the scarcity of skilled labor hinder progress in the field. Currently, the supply of AI personnel in Vietnam meets only 10 per cent of the domestic market’s recruitment demand.

Moreover, generative AI’s reliance on substantial amounts of high-quality data poses a significant obstacle, as data completeness, consistency, and accuracy impact model reliability and transparency. Strict data privacy and security regulations limit the data volume used for training generative AI models, making them susceptible to cyber-attacks and limiting their full potential. Inaccuracies or biases in training data can be amplified by generative AI models, leading to sub-optimal outcomes.

Layered infrastructure poses another challenge for generative AI, since generative AI heavily relies on databases. However, bank data and confidential information are often subject to limited access, making it impossible for AI to perform payment tasks related to customer information and confidential information.

For greater AI integration in the future, the development of large and high-quality data becomes essential in the banking industry.

To facilitate complex tasks related to customer information, security, and seamless financial transactions, it is imperative to continue researching and developing unified AI infrastructure solutions. This would enable AI to access vital information and execute automated processes without the need for constant surveillance.

Ideally, AI will be integrated with other breakthrough digital technologies like blockchain, which offers a highly secure database for data transmission and storage, ensuring both security and transparency, and facilitating the creation of interbank databases.

Furthermore, Vietnam’s AI development ecosystem and supportive policies still need substantial growth to catch up with other countries in Asia. Strategic investments in technology infrastructure, resources, and talent, including data scientists and machine learning experts, are critical for banks to retain competitiveness and stay prepared for emerging trends.