AI Can Reshape Insight Delivery and Decision-making — Visa Global Head of Risk Analytics and Insights

(US & Canada) Heather Tubbs, Global Head of Risk Analytics and Insights at Visa, speaks with Amy McNee, SVP, Solutions Architecture and Technical GTM, Informatica, in a video interview, about the scope of data and AI at Visa, AI challenges in the risk and fraud domain, the role of data literacy in driving AI strategy, and Visa’s approach to ethical use of AI.

Visa is a trusted network and world leader in digital payments, working to remove barriers and connect more people to the global economy.

With an undergraduate degree in psychology and an MBA, Tubbs started her career in a healthcare organization to help with strategic insights. The role led her into the data world, as she served close to a decade in healthcare analytics and insights, ensuring effective operations.

Next, Tubbs pivoted into the payment space at Visa, where she started as the Director of Insights and Analytical Delivery for North America. Then, she transferred to the global risk organization, focusing on new payment flows and regional analytics, and has recently taken over as head of the team.

Exploring the scope of data and AI at Visa, Tubbs stresses the criticality of AI in today’s business landscape. She notes that AI has the potential to significantly reshape insight delivery and decision-making in business processes.

Delving further, Tubbs states that her team is focused on not just creating data models but also providing insight delivery and decision support to executive leadership. When it comes to the evolution of AI, she refers to predictive analytics.

With AI’s capability to scrutinize historical data, businesses can predict future trends, says Tubbs. Especially in the risk and fraud domain, organizations can create a proactive approach to make more informed decisions and fight fraud in the ecosystem.

Further, AI aids automation, which allows employees to concentrate more on decision-making needs, thereby enhancing decision quality. This happens because people are more invested in what data means and less in data engineering or data capabilities.

Moving on to risk, Tubbs shares that AI plays a pivotal role in the organizational risk mitigation strategy. With AI, the organization can identify potential risks and propose countermeasures that can significantly contribute to business stability.

Therefore, Visa can be proactive in fighting fraud and risks, specifically in the payment landscape. Another usage of AI at Visa is in making real-time decisions with real-time analytics.

Given the billions of transactions a month, real-time analytics enable the organization to comprehend what the transactions mean and how to make prompt decisions around anomalous behavior. AI also fosters collaboration in the ecosystem and organization by encouraging different teams to work towards a shared objective.

Summing up, she refers to the cost-saving aspect of AI and maintains that Visa is driven to automate processes that have taken a significant amount of time historically.

Shifting to the other side of good AI, Tubbs affirms that AI can also be used by fraudsters for nefarious reasons. To avoid that, Visa constantly evaluates its models and algorithms. She notes that Visa has a dedicated team to look into the dark web to understand the actions of fraudsters.

Many a time, not just the cardholders but the merchants are targeted as well. So, it boils down to having a multi-pronged approach to stay ahead and looking at real-time data to assess anomalous behavior much quicker than before.

Shedding light on the role of data literacy in driving AI strategy, Tubbs believes that data literacy is crucial for an organization as AI evolves. She adds that it is a must for teams to know the meaning of data and not be limited to its whereabouts.

Otherwise, the team just becomes a reporting function devoid of insights. Thus, data literacy allows her team to better understand the nuances, specifically while building out machine learning models or designing effective AI algorithms.

Without data literacy, there will be more rework, making it take longer to deploy an AI model that solves business needs, says Tubbs. She states that as a data analytics shop, they partner with stakeholders who rely on them to understand how data works, which requires the team to have business understanding.

When asked how Visa approaches the ethical use of AI, Tubbs confirms prioritizing ethical considerations while leveraging AI or building AI models. Ethics is the backbone of Visa’s design thinking, she adds.

Even from a GenAI perspective, the organization ensures the inclusion of ethical components from the get-go. She explains the analogy of building the security system of a house before building the house versus after the house is robbed.

Furthermore, Tubbs states that Visa aims to ensure that AI models are transparent and completely understandable.

In conclusion, she reflects on how this ties back to data literacy—being able to explain what models are doing and articulate model decision-making for explaining to business stakeholders.

CDO Magazine appreciates Heather Tubbs for sharing her insights with our global community.