BS BFSI Insight Summit: 'AI's RoI may be slow, but governance can't wait'

The panelists said that re-skilling will be key to ensuring the workforce remains employable even as disruptive technologies continue to enter the sector

The return on investments (ROI) in technologies such as artificial intelligence (AI) will take time to materialise, and the immediate priority should be to increase spending on building governance frameworks around AI, top bank executives said at the Business Standard BFSI Insight Summit 2025. 

 

Currently, there is scope to utilise AI and automation across use cases such as customer service, underwriting, debt recovery, among other spaces to improve efficiency of banking services. 

 

Meanwhile, there should be valid grounds for utilising data and AI, and explainability of those aspects should be at the centre of it, executives added. 

 

Executives were in conversation with Nivedita Mookerji at a panel discussion titled ‘Impact of tech & AI on retail banking – Pros and Cons’ in Mumbai. 

 

 

“ROI is use case-by-use case. ROI is in terms of the overall philosophy of the transformative power, whether it is AI, Gen AI (generative AI), or tech initiatives that we are able to bring in,” said Arvind Vohra, Group Head, Retail Assets, Rural Banking Business and Microfinance, HDFC Bank. 

 

Governance frameworks around AI should be a core area of focus, executives concurred, with a need for further investment to build ethical standards around it. 

 

“Doing great work from a return perspective (ROI) is one angle, but investing into the tech to make sure the governance is right is equally important. It does need sizable investments on building the right data culture,” said Ambuj Chandna, Managing Director (MD) & Head, DBS Bank India

 

The acceleration of governance frameworks will be further enabled by right frameworks such as explainability of models, and their consistency. 

 

“The RBI (Reserve Bank of India) is very clear on this, that what you cannot explain, you cannot use it. It cannot be a black box. It has to have predictability, consistency. It has to have an explainability of what you are doing. Because it could cause a big loss or a reputational risk to the bank or the institution,” said Sachin Seth, Regional Managing Director, India and South Asia, CRIF. 

 

Appropriate governance frameworks would further provide ground to create trust in the ecosystem. This assumes importance since trust of the customer is the cornerstone of the banking ecosystem in the country. 

 

“Customer service is a place where the first trust for the bank gets created. And unless we get that, the use of technology will necessarily not mean that you are winning the trust of the customer. How responsibly the banks use AI is going to be the decisive factor as we move forward,” said Virat Diwanji, National Head - Consumer Banking, Federal Bank. 

 

The growth in AI-led technologies is further expected to accelerate the synthesisation of available data. 

 

“The data which is available with the bank currently is in structured format, it is in silos. With the technology and with AI coming in, there will be a kind of a unification of this. There will be a synthesization of the data which is available, cleaning of the data, which will enable these technologies to be far more efficient going forward,” Sanjay Mudaliar, ED, Bank of Baroda. 

 

The panelists added that re-skilling will be key to ensuring the workforce remains employable even as disruptive technologies continue to enter the sector. 

 

The banking sector has seen phases from the dawn of the internet to AI, and new jobs were created to sustain the ecosystem’s employability, they added. 

 

“We have had the internet, mobile banking, ATMs, core banking coming in. I think jobs never get lost, per se. Jobs get innovated, and new jobs get created,” Seth from CRIF, added. 

 

In terms of use cases where AI can be employed, customer services has taken the centre stage. 

 

“All our call centers are powered by AI-based transcribers. Any call that is happening is getting transcribed in real time. There is a quality check which is happening on the call to figure out whether it's a regular call or it's a complaint, what is the nature of the call that comes in there,” Chandna from DBS Bank India, said. 

 

He added that the text of these transcripts can be further mined to get intelligence from the product and policy perspective.