Case study: State Farm

With more than 58,000 employees across the board, State Farm is one of the largest insurance companies currently in the US. As a leading insurance brand, State Farm focuses on building brand relationships and trust with its customers."Quality service and relationships, mutual trust, integrity, and financial strength" are State Farm's core values. Additionally, these values outline the guiding principles for maintaining the organisation's success and keeping them on track.

Car accident claims may be classified as "total losses" when they are deemed beyond repair or where the cost of repairing the vehicle exceeds the value of the car. State Farm created an AI model called the Dynamic Vehicle Assessment Model (DVAM) to help predict a total loss at the beginning of the claim process, thus reducing the time required to settle the claim.

DVAM enabled State Farm to:

• Simplify, expedite and improve the customer experience.

• Reduce the total loss process from as high as 15 days to as little as 30 minutes.

Consequently, employees and agents have more time to focus on providing exceptional customer service.

(State Farm, n.d.)

By implementing DVAM, State Farm achieved the expected results, which helped strengthen alignment with the company's mission and values. DVAM was successful thanks to the comprehensive departmental collaboration and a thoroughly established AI governance system. The AI governance system at State Farm is comprised of the following:

• A chief data and analytics officer has primary executive accountability for responsible AI across the organisation and oversees the evolution of AI principles across the enterprise.

• A validation team reports to the chief data and analytics officer's department. The validation team oversees model validation and AI in software reviews and assesses various aspects of AI models.

• A risk and governance committee has members who come from various business areas, including law, risk management, privacy, IT, procurement, and executive partners. In addition to reviewing and approving model risk management procedures and guidelines, the committee provides strategic direction to the validation team.

The governance system ensured that DVAM was ready and met all guiding principles by launching in phases and thoroughly assessing the model before fully implementing it. Through the different phases, the governance process was transparent to all participants, preventing any surprises at a later stage. A control and oversight approach was applied throughout the lifecycle of the technology and accountability was assigned to the above groups.