The benefits of AI in the banking sector are already well recognised in areas such as improving customer experience, boosting efficiency and automating processes. However, the applications in AI are still evolving and have yet to reach the full potential. How else can AI transform banking?
ITNews Asia gets the lowdown from Sophie Dionnet, Global Vice President, Product and Solutions, Dataiku, and discusses how significant the impact of AI will be.
iTNews Asia: How is AI changing the financial industry in APAC?
AI is revolutionising how banks and financial institutions across Asia Pacific operate. Take customer service, for example. The days of waiting on hold to resolve a simple query are fading, thanks to AI-powered chatbots and virtual assistants like DBS Bank’s Digibank. These tools have dramatically reduced customer friction by offering instant, 24/7 support, whether it’s checking account balances or facilitating transactions.
AI’s impact on financial services, however, extends far beyond marketing and customer service and plays a vital role in risk management. For banks and other financial institutions, doing risk modelling is a core activity. With machine learning, they have an opportunity to augment risk modelling by incorporating newer risk factors in their approaches, uncovering new patterns, and be in a position to make better risk-adjusted business decisions on critical activities such as credit allocation or Know Your Customer (KYC). These advanced capabilities are crucial in a digital finance environment to safeguard both institutions and their customers.
The financial industry’s perception of AI is also evolving. In Singapore, research indicates that in 2023 alone, AI and machine learning initiatives at DBS Bank contributed approximately US$ 275.9 million (S$ 370 million) in incremental economic value through revenue growth and cost savings. A recent study by Accenture supports this, predicting that banks can boost their productivity by as much as 30 percent using generative AI over the next three years.
Materialising these efficiency gains takes different shapes and forms: from tedious analytics production acceleration down to financial or compliance reporting Gen AI-assisted production, opportunities for banks to streamline their day-to-day activities are massive, with the potential for a significantly reduced cost base down the line.
iTNews Asia: Banks very much started using AI in areas such as automation and data analysis in areas such as customer support in retail banking, sales and marketing. Chatbots, for example, are now a staple in customer service. With the increasing use of generative AI, what further financial and banking services are being transformed? How is the customer experience being reshaped?
While the full potential of Generative AI in banking is still unfolding, its immediate applications are already making a significant impact, particularly in areas where banks have already begun leveraging AI, such as automation and data analysis in customer support, sales, and marketing. In fact, Deloitte predicts that the top 14 global investment banks can boost their front-office productivity by as much as 27 to 35 percent using Generative AI.
Investment banks have long looked to streamline front-office tasks. While the concept of a fully autonomous “robo-banker” is not yet feasible, current AI applications can significantly enhance workflows. For instance, Generative AI can make financial documents more accessible by providing queryable formats, allowing junior bankers to retrieve relevant information quickly.
Additionally, Generative AI can assist in scanning internal deal documents and pitch decks, reducing research time and improving overall productivity. These improvements not only streamline workflows but enhance the quality of life for junior staff, who often bear the brunt of time-intensive tasks.
Generative AI also presents opportunities for improving CRM systems. Banks currently use predictive modelling to identify clients likely to undergo significant corporate events. By incorporating Natural Language Processing (NLP) features derived from news and management commentary, AI can improve the accuracy of predictive models, enabling bankers to identify high-value clients and tailor their outreach strategies more effectively.
Overall, the technology’s adoption must be accompanied by rigorous risk management and regulatory frameworks to ensure its safe, accurate, and effective use.
iTNews Asia: Banks have traditionally prioritised security, process organization and risk management. Can the use of AI deliver real value, for example in risk assessment or fraud prevention? What are the future possibilities?
One of the most significant opportunities AI presents for banks is the ability to process vast amounts of data in real time. This capability allows banks to gain deep insights into customer behaviour, identify potential risks, and detect fraudulent activities. AI-driven algorithms can analyse transaction patterns, spot anomalies, and trigger alerts for suspicious activity, enabling banks to proactively safeguard their customers and assets.
For example, BGL BNP Paribas looked to strengthen key risk control processes through advanced analytics. Although the bank already had a machine learning model in place for advanced fraud detection, with limited visibility and data science resources, the model remained largely static.
When changing the model, the challenge was to harness a data-driven approach across all parts of the organisation. This initiative brought together data analysts and business users from the fraud department, along with data scientists from BGL BNP Paribas’ data lab and Dataiku. Through this collaboration, the bank successfully developed a new fraud detection prototype that delivered clear business value.
Ultimately, as AI technology continues to evolve, the possibility of its application to traditional banking will remain vast. By embracing AI, traditional banks can position themselves for long-term success in an increasingly competitive and dynamic landscape.
iTNews Asia: What do you see as the challenges of implementing AI in banking? Do you see issues in governance, transparency in the use of data, data privacy? How can we overcome them?
One of the primary concerns is the complexity of AI models, which can often make it difficult to understand how they arrive at their predictions. AI models often operate as “black boxes,” making it difficult to explain their decision-making processes and leading to unintended biases. The usage of LLMs is only amplifying these concerns.
If built without the right explainability, AI use cases cannot be applied to critical banking activities, notably as they would be in no position to pass audits from regulators or simply would fail to be used by the business functions for lack of trust and maintainability.
On the other end of the spectrum, banking decisions are largely supported by large volumes of analysis usually done in outdated tools like spreadsheets, including for critical analytics. These legacy systems, while familiar, hinder growth and efficiency, ultimately creating a compliance burden that will need to be addressed down the line.
Overcoming these (legacy) challenges requires a multi-faceted approach. Banks must invest in developing AI expertise among their staff, including risk managers and compliance professionals. Additionally, implementing comprehensive governance structures that cover the entire AI lifecycle—from development to deployment and monitoring—is essential to embed rigorous risk management practices in AI system building.
– Sophie Dionnet, Global Vice President, Product and Solutions, Dataiku
Advanced platforms provide a centralised, governed environment for data processing and model development, reducing barriers to governance enforcement and augmenting compliance readiness. Ensuring there is no gap between data pipelining and modelling also plays a critical role in improving system accuracy and auditability.
iTNews Asia: Not all banks in APAC are embracing AI at the same speed – a lot depends on their data maturity and AI readiness. How can they prepare? What should banks do to get themselves AI-ready?
To put it simply, a tailored approach is essential. The crucial first step is to assess the bank’s current capabilities and challenges. This involves evaluating existing data infrastructure, identifying the level of AI understanding and adoption, and pinpointing specific business problems that AI could solve. This assessment will provide a foundation for developing a tailored roadmap for AI adoption.
Improving data foundations is an important part of this process – this is essentially the fuel that powers AI models and enables accurate predictions and insights. Banks must invest in improving data quality, centralising data management, and upgrading data infrastructure to ensure that data is accessible, reliable, and suitable for AI use.
Concurrently, developing AI skills and expertise to develop, implement, and maintain AI solutions is essential. This can be achieved through hiring or training skilled individuals, partnering with external providers, and providing ongoing training opportunities.
Ultimately, choosing the right AI platform is a critical decision and will be essential to maximising the benefits of adoption. Banks should seek a balance between high power and low risk. Platforms like Dataiku allow users to break away from spreadsheets, build intelligent and potent (often machine learning) models, visualise their data, and build value-adding insights swiftly and without friction. These features allow stakeholders across the firm to work collaboratively on the same projects with clear, governable, and auditable oversight every step of the way, minimising risk while retaining power.