Report: Health Insurance Artificial Intelligence/Machine Learning Survey ...

The AI Pulse Check: 84% of Health Insurers Are Now Using Machine Learning, But Regulators Are Watching

The integration of Artificial Intelligence (AI) and Machine Learning (ML) is rapidly reshaping the insurance landscape, particularly in the complex sector of health coverage. To keep pace with this transformation—and to ensure consumer protections remain paramount—the National Association of Insurance Commissioners (NAIC) recently completed a massive study capturing the current state of AI adoption among major health insurers.

The resulting report, "Health Insurance Artificial Intelligence/Machine Learning Survey Results," is critical reading for industry professionals, regulators, and tech enthusiasts alike. Conducted across 16 states and polling 93 major companies, this survey provides the first comprehensive look at how health carriers are deploying AI, managing third-party risks, and building governance frameworks to meet ethical standards.

This report confirms that AI is no longer an emerging technology in healthcare; it is deeply embedded in core business functions, from utilization management to fraud detection. The data gathered will directly inform the next steps for regulatory oversight, making these findings essential for understanding the future compliance environment.

Key Takeaways from the NAIC Report

• Overwhelming Adoption Rate: 84% of health insurers surveyed are currently using AI/ML across various product lines, including Individual Major Medical, Group Major Medical, and Student Plans.
• Focus on Efficiency and Cost Control: The primary applications of AI are centered on operational efficiency. Utilization Management practices (71% adoption) and Prior Authorization for approval (68% adoption) are leading the charge, closely followed by Disease Management programs (61%).
• Hybrid Development Model Dominates: While only 10% of companies develop AI solutions entirely internally, the vast majority (55%) use a hybrid approach—developing internally while incorporating third-party AI/ML components. A significant 15% rely entirely on third-party solutions.
• Cautious Approach to High-Risk Decisions: Insurers show restraint in using AI for sensitive consumer actions. Only 12% of companies use AI/ML to review prior authorizations for denial, and only 14% use AI to infer sensitive data, such as race.
• Governance Is a Priority: Companies are proactively adopting principles focusing on accountability, transparency, security, and privacy. Common testing methods include exploratory data analysis (EDA), cross-validation, and conducting explicit equity and compliance audits to detect bias and drift.

The State of AI Deployment in Health Insurance Operations

The survey dissected AI usage across several key operational domains, revealing which areas are fully mature and which are still in the planning stages (Table 5).

Operational Areas Leading Production Deployment:

The data shows a clear preference for implementing AI in areas that impact strategic decision-making and high-volume transactions:

• Strategic Operations: 52 companies reported AI already in production.
• Utilization/Severity/Quality Management: 42 companies reported AI already in production.
• Fraud Detection: 42 companies reported AI already in production.
• Sales & Marketing: 37 companies reported AI already in production, often leveraging the technology to enhance quoting, online sales, and the overall customer shopping experience.

While these efficiency-focused areas are mature, the report highlights that deployment is still growing in traditionally complex areas like Product Pricing and Plan Design (22 companies in production) and Prior Authorization (18 companies in production). Notably, 16 companies plan to deploy AI in Prior Authorization within the next year, indicating rapid acceleration in this critical consumer-facing function.

Addressing the Third-Party Ecosystem

One of the NAIC’s core objectives was to understand the reliance on external vendors. The finding that 55% of insurers integrate third-party components into their internally developed systems underscores the complexity of the current AI supply chain. Regulators are keen to understand how insurers maintain oversight and accountability when the models and data they rely on are sourced externally.

Barriers to Adoption

For the 16% of companies not currently using AI, the reasons cited largely revolve around strategy and infrastructure, rather than outright rejection. The top barriers included "No compelling business reason at this time" (20%), closely followed by challenges related to "Lack of resources and expertise" and "Reliance on legacy systems" (both 13%). A small percentage (8%) are waiting for clearer regulatory guidance before committing to major AI investments.

The Regulatory Challenge: Governance and Accountability

The survey results on governance demonstrate that insurers are actively working to align with the NAIC's foundational AI Principles, particularly those related to accountability, transparency, and security.

Insurers reported adopting practices to manage the risk of unintended consequences, focusing on:

• Bias and Unfair Discrimination Testing: Companies routinely use cross validation, performance metric tracking (like AUC and F-score), and, crucially, specific equity and compliance audits.
• Transparency & Disclosure: While governance practices are strong, the transparency metrics show room for growth. Companies were asked about practices related to disclosing non-FCRA data collection and providing consumers with a chance to correct that data. This area will likely be a focal point for future regulatory guidance.
• Human Oversight: A key safety net identified is human intervention. Insurers reported using hybrid models where AI supports, but does not fully automate, critical decision-making processes, particularly in highly regulated functions.

Conclusion

The NAIC’s comprehensive survey confirms that the health insurance industry has enthusiastically embraced AI/ML technologies to drive efficiency and optimize outcomes. However, this growth comes with significant regulatory responsibility. While insurers are taking steps to govern AI usage and test for bias, regulators now have crucial data to develop targeted frameworks that ensure these powerful technologies are used ethically and responsibly, protecting consumers without stifling innovation.

The path forward involves continuous monitoring and the refinement of standards to match the velocity of technological change. This report is the essential starting point for that process.

Want to dive into the full data set and analysis? Download the complete NAIC Staff Report today: Download the Full Health Insurance AI/ML Survey Results Report