Payments 3.0: For AI, Have a Strategy First

Success with artificial intelligence is about strategy, but most conversations about AI today focus on tactics.

Although financial services have used AI for decades, the growth of generative AI has led to a broader set of use cases. Innovators are looking for ways to apply artificial intelligence to everything from customer service to regulatory compliance.

While each of these tools can be valuable, the greatest success will come to those companies that use AI for strategic goals.

Last month, I attended the AI-Native Banking and Fintech conference organized by Spring Labs, an AI company focused on customer support and compliance. The sessions focused on AI for functions like customer service and process automation, along with discussions on the regulatory implications of using AI.

In one panel, Jordan Wright, the chief executive and co-founder of Atomic, a company focused on API account connections, bridged the discussion from tactical to strategic. He described how his company uses AI for things like developing sales pitches and offering account-management tools.

Wright said he hoped AI would enable Atomic to grow to a $1-billion company, with the 20 employees it has, by helping it offer additional products and services while operating efficiently. This framing shows how AI can apply to a larger strategy. While AI can spot fraud, reduce costs, or simplify document reviews, no company will have a monopoly on it.

Companies must plan how to differentiate themselves when all their competitors also have chatbots for customer service, machine learning for identifying fraud, and generative AI for marketing.

Margaret Mayer, chief technology officer at Zions Bank, predicted that, over the next year, banks will see incremental gains from AI. But in five years, she said, AI will transform the industry. She said her bank is preparing by having a data-science team and an internal sandbox to test products.

Still, she worried about the risks posed by AI. One concern she cited was how well customers would understand what they are consenting to when they give permissions for systems to access data, particularly in an open-banking environment.

A second concern Mayer and others raised: How regulators might react to AI-driven changes in the industry.

One answer is that banks are already prepared to deal with regulatory risks. Anne Romatowski, a markets-program manager for the Consumer Financial Protection Bureau, pointed out that current laws do not change just because there is new technology.

Similarly, customers’ core needs do not change just because the technology to meet those needs has changed. They still need to make payments, save money, borrow money, and manage their financial lives. Companies will need to keep customer needs centered as they begin to integrate new tools.

An additional strategic consideration companies will need to consider is that two things limit the power of AI tools: the robustness of the models and the quality of data used to train those models.

As Derek Higginbotham, president and chief executive of First Electronic Bank pointed out, the models can drift over time based on the data they receive, and that can lead to problems if management is not paying attention.

The second risk is the “black swan” event that fits a model or its training data set but turns a portfolio upside down. Companies will need to think holistically so they know when to adjust a model, or even go against an AI system’s recommendation in response to novel market conditions.

All of this requires that a company’s leadership knows what its core business is and what its core strategy is. Without that, AI is just a tool to manage a bureaucracy.

—By Ben Jackson bjackson@ipa.org