The decisions we make every day have a significant impact on customer experience, performance and profitability. In the insurance industry, these decisions are often amplified by our imperfect biases, specifically in claims decisions and settlements.
Embedded Claims Bias
Claims bias may exist in the work that file handlers perform daily. According to a write-up by Eagle International Associates, the insurance industry is experiencing a threat of “implicit bias” from claims professionals and legal teams. Implicit biases are potentially costing the insurance industry millions and could bring up numerous claims of bad faith against carriers.
The evaluation and recommendations made by both claim professionals and lawyers often have a significant impact on settlement and jury verdicts of court cases. This is particularly troublesome as countries around the world adopt a more progressive view of diversity and human value. Subsequently, a pattern of implicit biases could result in allegations of discrimination against insurance companies.
Biases in AI algorithms and models are complicating the issue. A PropertyCasualty360 article discusses the problem with AI systems in underwriting, claims processing and fraud. The inputs and outputs to an AI model are not always transparent and often they are constructed from potentially biased data. In turn, the algorithm may be based upon data-based biases and historical data rather than unbiased data sources.
Solving Your Biases with AI
The goal is to remove discrimination from the claims handling process. This is important because claimants and claim outcomes can be unintentionally prejudiced.
The insurance industry is already finding ways to combat inequality and biases in their AI algorithms and models. Traditionally, factors such as credit score, income level and zip codes can discriminate against minorities and disadvantaged individuals. According to Digital Insurance, insurers have been working on leveraging smartphones and sensors in order to quantify risk in a more personalized manner. Therefore, insurers can price premiums towards the individual rather than towards a group.
Insurance AI has also been effective at recognizing racial and ethnic biases. Systems such as IBM’s AI Fairness 360 work to detect and remove biases by giving users metrics, models and algorithms for machine-learning models. This is an important tool for the industry going forward as claims processing becomes increasingly automated.
Addressing Inherent Biases
AI solutions are an important first step towards eliminating claims bias. However, identifying and auditing your inherent biases are the most important steps you can take to protect yourself and your company from legal action.
According to the Journal of Accountancy, creating a mental model is one way to identify and address inherent biases. This process allows a person to consider different variables associated with a situation before making a biased assumption towards the person involved.
The Bottom Line
The bottom line is that claim bias is an inherent issue for all of us in the insurance industry. To avoid these implicit biases, it’s important to continuously raise awareness throughout your team and put processes in place to identify, audit and address them.
Want to keep up with the latest trends in claims? Subscribe to receive the VCA Software e-newsletter.
