Retail banking: AI is changing the world, responsible adoption is crucial

Retail bankers say AI can reshape banking, but trust, governance, explainable models, and high-quality data are essential for safe and sustainable adoption

Artificial intelligence (AI) can transform banking, but retail bankers say there are caveats. At a panel discussion moderated by Nivedita Mookerji at the Business Standard BFSI Insight Summit 2025, industry leaders — Ambuj Chandna, MD and head- consumer banking DBS Bank India; Virat Diwanji, national head – consumer banking, Federal Bank; Sanjay Mudaliar, executive director, Bank of Baroda; Sachin Seth, regional managing director – India & South Asia, CRIF High Mark; and Arvind Vohra, group head – retail assets, HDFC Bank — highlight that trust, strong governance, explainable models, and high-quality data are critical. Edited excerpts:

 

What’s happening on the AI front, and what have been the initial customer benefits at DBS Bank? 

AMBUJ CHANDNA: For me, this journey began with machine learning, evolved into AI, and is now rapidly moving toward generative AI. Across the entire customer lifecycle, the impact has been significant. At the acquisition stage, for instance, our propensity models, customer nudges, and engagement strategies have become far more precise. By using both structured and unstructured data, we can now engage customers in a more targeted and contextual manner, responding to their needs at the right moment. AI is also transforming core lending functions — underwriting, credit decisioning, and even debt recovery. Earlier, calls were largely recorded and reviewed manually. Today, our call centres are powered by AI-based real-time transcribers. What gets stored is no longer just a voice recording, but structured text data. This allows us to mine conversations for insights. A significant part of banking and financial services is already being reshaped by AI, and this will only deepen. 

 

Will the world of tomorrow be very different in terms of customer experience? What are the challenges? 

VIRAT DIWANJI: The world ahead will certainly be very different because of how deeply AI and technology are penetrating banking. Traditionally, bankers have been very good at analysing numbers. AI, however, is a double-edged sword. Used well, it can deliver enormous efficiency and speed; used irresponsibly, it can introduce bias and create new risks. That is why responsible adoption is critical. Banks cannot ignore AI. From a customer service perspective, this is where trust in a bank is first built. Technology alone does not create trust; how responsibly it is deployed does. AI has already transformed this space through chatbots, voice assistants, and real-time analytics. These tools handle large call volumes, offer multilingual support, and make service more inclusive and accessible. 

Going forward, the winning formula for customer experience will be simple: AI for speed, humans for empathy, and data for trust. That balance will define how banks build lasting relationships. 

How is Bank of Baroda ensuring customer trust and security in digital banking? 

SANJAY MUDALIAR: Trust is the foundation of banking, and it is the banker’s duty to ensure customers feel secure and comfortable. Customer trust depends on how they are received, serviced, and how grievances are addressed over time. Technology, including AI and GenAI, can significantly enhance this by improving efficiency, speed, and consistency across touchpoints. Another critical factor is the information shared with customers. Much of bank data today is siloed and unstructured. AI can unify, clean, and synthesise this data, making services more reliable, personalised, and effective. 

 

How can AI improve creditworthiness assessment? 

SACHIN SETH: Lending is the most profitable franchise for banks, and credit decisions are central to it. AI has significantly evolved in this space, and the RBI’s recent Ethical AI framework provides guidance for responsible AI use — not just in lending, but across banking operations. 

AI is enhancing underwriting through faster turnaround time, higher accuracy, and the ability to leverage surrogate and multidimensional data. For example, credit bureau models can differentiate between a 100cc bike and a 350cc premium bike based on usage patterns, income sources, or location — something impossible to do manually. We build thousands of models to optimise approval rates and reduce defaults. AI cannot be a black box. Models must be predictable, consistent, and explainable to avoid reputational and financial risks. This journey in credit underwriting is ongoing and will evolve over the next few years. Finally, the account aggregator framework is gaining traction. While usage is currently around 3%, it is growing at 20% month-on-month. This data enables better spend analysis, risk assessment, fraud detection, and income evaluation, strengthening credit decisions further.

 

Are boardrooms now discussing AI more actively? 

ARVIND VOHRA: Transformational changes like AI inevitably reach boardrooms — across banks, regulators, and stakeholders. Our belief is clear: as innovation cycles shorten, responsible AI must be anchored in strong governance, ethical design, transparency, and accountability. These pillars are essential to ensure AI builds trust rather than erodes it.

 

Is Return on Investment (RoI) a concern with AI investments? 

VOHRA: RoI from AI and GenAI is inherently use-case specific. While finance teams will rightly evaluate returns, the real value lies in the broader transformation of operations and customer engagement. 

AI’s power lies in improving relevance — using micro-segmentation and propensity models to deliver the right offer to the right customer at the right time, while respecting privacy and consent. 

Beyond sales, GenAI can dramatically improve internal processes by scanning policies, emails, and customer queries using LLMs to surface precise guidance. 

CHANDNA: While RoI is important, investing in the right foundations is equally critical. Strong data quality, security, access controls, and ownership are essential — after all, it’s garbage in, garbage out.

Model governance is just as important. Building models is easy; ensuring they are fair, explainable, and well-tested is not. Institutions must embed diligence, fairness, and accountability into AI deployment.

 

Will AI take away jobs? 

Seth: AI is just another wave of change—like core banking, ATMs, or mobile banking. Jobs do not disappear; they evolve. New roles emerge, skills change, and productivity improves. The real challenge is reskilling and adaptation. Boards are not afraid of AI; they are uncertain about where to invest. 

 

How can AI be a true differentiator? 

MUDALIAR: AI must first be anchored in strong governance, clear policies, and oversight. Customer service and fraud monitoring are currently the most common use cases, but the scope is limitless. RoI will follow as efficiency, relevance, and service quality improve over time.

 

Will AI deliver on its promise? 

DIWANJI: Scepticism is natural with any hyped technology. AI is a journey, not a one-time bet. The key is to invest based on business needs, not fashion. ROI today is incremental and use-case driven. Over time, as AI matures and scales, the returns will become clearer.