On a recent SRM Perspectives Live! webinar, it was noted that more than 80% of financial institutions are actively engaged in artificial intelligence (AI) initiatives, with the remainder in the planning stages. This is a remarkable shift for a single year. As recently as early 2024, many banks and credit unions remained uncomfortable with the idea of deploying AI.
The first signs of a shifting tide appeared in the back office, where employees were able to experience efficiency gains firsthand without courting the unknowns of directly exposing clients to AI technology.
Consumers are gradually becoming acclimated to AI models as well, prompting more institutions to explore customer-facing applications. The rapid growth of AI agents across sectors beyond banking is likely to accelerate such use cases.
Friction on Demand
The pace at which AI surpasses human performance at a variety of tasks is accelerating. Handwriting and speech recognition took nearly a decade to achieve human parity. By the time image recognition was launched in 2009, that time had been cut in half. Now tasks like language translation and code generation are going from zero to 100 in roughly a year.
Language understanding (including translation to/from non-English) and “common sense completion” are among the most recent breakthroughs, both of which hold great promise for customer service. On the latter, AI agents can answer a greater number of non-standard inquiries by knowing where to find relevant data across disparate systems. They’re already smart enough to schedule appointments (by phone or in-branch) for any necessary follow-up after providing a degree of near-term closure.
State-of-the-art systems enable each financial institution to customize its AI guardrails, defining the level of data agents are permitted to access from core systems. It’s also capable of adding friction in real time in heightened fraud risk scenarios.
In a separate poll, a majority of financial institutions indicated they believe the regulatory stance of the new administration will have a positive impact on AI advancement. An early post on Whitegouse.gov, launched way back during the Clinton Administration, offered their “Ten Rules of the Internet.” The first was “The private sector shall lead the Internet.” The current policy extends that philosophy to AI.
Giving Bots a Human Face (and Voice)
Research has shown that attaching facial features to a machine, or designing it to resemble a household pet, is sufficient to foster a level of human connection, causing people to relax their misgivings. Ashish Garg, the co-founder and CEO of AI firm Eltropy who appeared on our SRM Perspectives Live! session, believes the injection of empathy into the vocal inflection of AI chatbots will have a similar effect in spurring adoption.
Such enhancements are already underway. Rather than the dispassionate monotone many associate with voice response units, newer systems can be made to sound less, well, robotic. This is a dynamic feedback loop, adjusting the AI agent’s tone in response to a reading (in word choice and tone) of the customer’s level of frustration, anger, and stress. Innovators and solutions providers like Sesame have made great strides in this area. Eltropy’s Garg suggests this capability may be particularly powerful in reframing debt collection from a dogged pursuit into an empathic, collaborative “self-cure” interaction.
Additionally, such agentic capabilities offer the promise of delivering valuable dashboard data in a snapshot morning report highlighting the most common questions posed during off-hours. For instance, has there been a significant reaction to some news story deserving of further staff education or proposed messaging? Is there evidence of a fraud trend? Do callers seem unusually angry or panicked?
The Bottom Line
We believe the risks posed by AI are manageable and the opportunities across the enterprise are too great to leave untapped. At a minimum, banks and credit unions should be encouraging their employees to familiarize themselves with the technology, while internal technology leaders should plan and pursue measurable pilot projects. To learn more about AI strategies for financial institutions, please contact Larry Pruss or Connor Heaton for a consultation.
Larry Pruss on May 12, 2025 9:30:00 AM
