Health plans have rapidly embraced AI for multiple use cases and leaders are focused on creating measurable performance improvement and achieving front-line buy-in.
To learn more about how health plans are leveraging AI to improve payment integrity and performance, Becker’s Healthcare spoke with Will O’Neill, vice president, healthcare solutions, and Aneesh Menon, assistant vice president, payment services at EXL.
Payers’ cost control focus
National health expenditures have risen sharply over the past few decades, growing faster than inflation. In 2000, health expenditures were about $1.4 trillion and have since more than tripled to $4.9 trillion. From 2022 to 2023, health spending grew 7.5% and is expected to continue to grow at a similar rate for the foreseeable future.
In addition to healthcare’s already high costs and the high rate of healthcare inflation, the U.S. Department of Health & Human Services Office of Inspector General estimates that healthcare fraud, waste, and abuse may account for up to 10% of expenditures in government programs. Studies have shown that fraud and waste could account for 10% to 25% of total healthcare spending.
EXL sees payers using AI to control both medical and administrative costs. By embedding AI into workflows, it can reduce tedious, repeatable work done by humans, improving payment integrity, accuracy, productivity and performance.
O’Neill believes that over the next five years, AI has the potential to dramatically reduce fraud, waste and abuse in healthcare, possibly by as much as 50%.
From pilots to performance
While health plans see the immense potential of AI to improve performance and payment integrity, many plans have only experimented with AI. O’Neill observed that experiments often use AI as a copilot. However, copilots don’t speed up or eliminate work and don’t improve performance.
For health plans to meaningfully improve productivity and make progress on cost control, they will need to pivot away from AI copilots toward agentic implementations. When deployed at scale within workflows, agents can drive measurable outcomes.
“When you incorporate agents into core workflows to automate repetitive tasks, you’re going to reduce the administrative burden and increase throughput,” said O’Neill.
Developing organizational capabilities
A misconception is that organizations can achieve their desired cost-control outcomes simply by implementing technology. Real-world experience has shown that success requires foundational capabilities and change management.
Because data is siloed and largely unstructured, health plans need improved data management and data engineering capabilities. In addition, it is not enough for AI to help make decisions; engineers and architects must be able to explain these decisions.
“We need to have architects that don’t just think about driving the outcome, but think about explaining the outcome,” O’Neill said. “Explainability needs to be incorporated into engineering practices.”
Also essential is developing strict governance practices guiding what AI and agents are allowed and not allowed to do.
Change management
Successfully deploying AI requires overcoming employee skepticism by building trust, engagement and buy-in.
“Change management is critical,” O’Neill said. “If you want your operators to embrace AI, you need to show them how AI is going to get rid of work they hate.”
Menon noted that when organizations fall short of achieving their AI objectives it is often due to human implementation, not technology limitations. One critical element to ensure technology is driving financial and organizational success is communicating to employees that AI helps and supports them, not replace them.
“You have to ensure the organization is adopting the changes you are making with AI; that’s where most organizations are failing,” said Menon. “They should feel like it’s a support system that will reduce friction.”
