Krish Swamy, Chief Data and Analytics Officer at Citizens Financial, speaks with Jonathan Steinert, Partner for Financial Services Technology at Guidehouse, about the principles driving organizational approach to AI, risk management, and upskilling, embracing AI thoughtfully, and the future of banking with AI.

Building trust and safety into AI: Data discipline and human oversight

Swamy shares his perspective on the fundamental principles driving his organization’s approach to AI, data privacy, and responsible technology use. He emphasizes that a foundational step to responsible AI and data usage is ensuring rigorous data discipline.

“A lot of that starts with having good data discipline in terms of how the data is organized, and how it is categorized.”

This categorization forms the basis of how data is ultimately used. According to Swamy, the citizens’ information classification policy lays out clear definitions for various data types:

• Public
• Internal Use
• Confidential
• Restricted

He adds that while confidential or restricted data could be used in a limited manner within the bank, it cannot be exposed outside the virtual private cloud.

Swamy stresses that data protection and privacy are central to everything they do.

The second core tenet he shares is the caution in the deployment of generative AI (GenAI).

“We are very deliberate about not putting anything out there that is generative and that goes directly in front of customers.”

He strongly advocates for a human-in-the-loop model, particularly due to the current limitations in GenAI’s performance and accuracy.

“We need to have a human in the loop whenever we start to use generative AI, especially because there are a lot of unknowns in terms of performance, quality, and accuracy that still exist with this technology.”

Until those unknowns are resolved, he believes a human oversight model is the safest course.

Risk management and upskilling

Swamy also highlights the importance of going beyond traditional model risk assessments. He lists a comprehensive risk lens that includes operational risk, reputation risk, and people risk.

The human factor, Swamy explains, is not just about oversight — it is also about capability. Swamy notes that it is critical to train humans to be effective users of the technology.

“The human being at a higher level of skill becomes important with everything that we are doing. That is also a big area of focus.”

For Swamy, the evolution toward AI integration is not about hurdling over barriers but about approaching the journey with care, intention, and realism. He emphasizes that the questions and challenges surrounding AI should be seen less as roadblocks and more as essential checkpoints that shape smarter decision-making.

Delving further, Swamy discusses a deliberate, incremental approach to transformation, grounded in practical learning. This philosophy applies most critically to data readiness. He explains that sometimes progress needs to pause so foundational issues, like the quality and structure of data, can be addressed first.

“It allows us to be more measured in terms of how we handle data, particularly. So I wouldn’t see any of these things as necessarily barriers.”

That said, Swamy acknowledges that change is required, and the rollout of AI solutions often reveals gaps that must be filled before moving forward.

Another critical insight he shares is the human side of transformation. “…users need a lot more enablement, encouragement, and training for them to be more effective users of these solutions,” he says.

Swamy sees this learning curve as something we all experience, even in our personal lives. “Using AI is something that I think all of us are still learning in our day-to-day lives.”

He reflects on his own evolving behavior when interacting with technology. This shift, he notes, came well after the introduction of ChatGPT. In Swamy’s view, human adoption takes time, and that must be allowed.

AI, data sophistication, and responsible transformation

Moving forward, Swamy shares his vision for how data, artificial intelligence, and evolving business expectations are reshaping the future of banking. He points out a clear trend across the banking industry: the need for elevated sophistication in how data is leveraged across all functions.

“I think one clear trend is that I think with all of our different product practices, the way we market, the way we service our customers, the way we do our pricing, and the way we detect fraud, I think we have to get even more sophisticated.”

He sees data analytics, machine learning, and AI as not just innovations but as inevitable tools that must be tightly integrated into everyday operations.

“Within Citizens, I look forward to us using that data foundation to build even more sophisticated solutions using machine learning and AI.”

Safety and responsibility at the core

While Swamy is optimistic about technological advancements, he is equally insistent on the need for responsible deployment. He likens the use of powerful AI tools to working with dangerous equipment in a workshop, emphasizing that the right safeguards are crucial.

As AI tools grow more powerful, safety, explainability, and risk governance must also grow in lockstep, says Swamy.

“The bar when it comes to safety, explainability, and risk governance is just going to keep rising as we get more sophisticated using these tools. We have to prepare for that, and we’ve had that pretty much as a day one expectation.”

Elevating human capability

In parallel with deploying AI solutions, Swamy stresses the importance of upskilling the workforce to better engage with these tools.

“Human users will need to get increasingly more comfortable with the use of some of these tools, and that’s going to require them to build their own business and technical skills to be able to ask the right questions.”

One of AI’s biggest lessons, he notes, is that the quality of the answer is highly dependent on asking the right questions, thereby shifting focus to the quality of input. Swamy adds, “That’s probably always been the case in the business world, but AI is bringing that to light.”

This shift will place a higher demand on employees to become more business savvy and more effective in how they interact with intelligent systems. He states, “I see that as a trend where the quality of users is going to continually rise in terms of their ability to use a lot of these tools.”

Rethinking the banking model

Finally, Swamy urges starting to look at a broad-scale transformation of how banking as an industry is practiced. He makes it clear that banking’s future a decade from now will be fundamentally different. This calls for leaders to revisit the core purpose of their services and challenge long-standing models.

Furthermore, he suggests understanding the customer standpoint and then meeting the need rather than optimizing existing processes.

Concluding, Swamy says, “It’s going to require us to rethink our business models and rethink the way we execute against customer expectations in a whole different way than what we’ve done before.”

CDO Magazine appreciates Krish Swamy for sharing his insights with our global community.