Valley Bank is exploring generative AI, expanding machine learning operations and paying keen attention to data throughout the bank’s operations. 

Chief Data and Analytics Officer Sanjay Sidhwani joined the $61 billion, Morristown, N.J.-based bank in April in the new position. He aims to make “Valley a more data-driven organization to use data insights and advanced analytics” to better serve customers and make data a market differentiator for the bank among its competitors, he told Bank Automation News.

“As the bank has grown at a fast pace over the last few years, as the industry has matured, it is almost table stakes to have deep insights about your customers to know what their needs are.”  

The financial services industry is evolving to make better use of data and insights in decisioning and Valley Bank believes data is going to be a key enabler in its growth, he added. 

Sidhwani was executive vice president of data and analytics at Santander Bank and executive vice president of business analytics at KeyBank before joining Valley, according to LinkedIn. 

BAN sat down with Sidhwani to understand how the bank is developing its data, AI and cloud strategy. What follows is an edited version of the conversation: 

Bank Automation News: How is Valley Bank using AI and what use cases of gen AI are being explored?  

Sanjay Sidhwani: The bank has machine learning techniques in place within risk management, financial crimes, fraud detection, marketing and credit processes. 

We do not currently have live use cases with generative AI, but we are actively exploring a few internal use cases including document summarization. 

At the bank, document summarization is being explored to create an internal chat agent for the bank’s entire knowledge base, internal policies, product documentation and system documentation. We want to create internal-facing chat agents that make it easier to access information. That will add more value to activities such as understanding customer needs. That’s the primary focus of the internal use of gen AI. 

BAN: What are the hurdles of implementing gen AI for consumer-facing operations? 

SS: Accuracy of the information. The way we use it would be through a large language model, but the model is built on primarily publicly available data. The hurdles would be how current that data is and whether the data is accurate or not. 

There’s a lot of open questions around what data has been used to train these generative AI models, a lot of questions on copyright and intellectual property. Additionally, the government is publishing new guidelines to keep up with.  

There’s a lot going on around gen AI, so I think putting it into customer-facing applications would be a bit risky at this point because we don’t want inaccurate information affecting our customer experience. 

BAN: What is the AI working group that Valley Bank has created?  

SS: This is a cross-functional group. It has representation from the top 15 people from all lines and functions of our business including retail, commercial, marketing, digital risk, information, security and legal compliance. 

This group’s main purpose is to develop and guide values of AI and build strategies for adoption of the technology. It’s essentially like an umbrella group that facilitates the awareness and implementation of AI across the organization. 

We have created a policy that further educates our associates on what AI is and how it should be used responsibly to provide oversight on the use cases. 

BAN: How does the bank approach building its own AI model versus looking through third-party vendors? 

SS: I think a lot of service providers are building AI into either machine learning predictive techniques or generative AI within their service offerings. 

We are exploring bringing on an enterprise LLM that can be trained on our data, but it must be commercially available. We’re not large enough or have the resources to build our own AI. That will be generally our path, to leverage commercially available products and then to customize them. 

BAN: How should FI leaders approach data and AI?

SS: Everybody needs to have a basic level of data literacy. It doesn’t matter if you are working in customer-facing operations, marketing or building financial products, everybody needs to have some level of data literacy. 

For people who are in the data space, learning data engineering skills is critical. If you remember, four years ago, all magazine covers said that the hottest job of the 21st century is going to be a data scientist. 

I think data engineers are the upcoming stars and they will be for a while because of the explosion of AI and I’m not saying that data scientists’ jobs are obsolete, but it’s been commoditized into something else. Having access to good quality data and being able to manipulate it, that is going to be a much tougher problem to solve and it’s very unique within each company. Each company’s data problems are different. People who have good data engineering skills and know how to wrangle data can increase the quality of the data and be more effective at their organizations.

Get ready for the Bank Automation Summit U.S. 2024 in Nashville on March 18-19! Discover the latest advancements in AI and automation in banking. Register now.