The 1940s marked the beginning of the use of Artificial Intelligence (“AI“) with the decoding of the Enigma machine following World War II, although its use in the insurance industry was only implemented during early 2010 to enhance efficiency and reduce costs by analysing vast quantities of data. It has since been employed in claims processing, customer service, and risk assessments. Despite the fact that insurers pay large sums in claims annually, a small percentage of claims are denied, primarily due to omitted or inaccurate information submitted during the underwriting process.
Underwriting is the standard process in the insurance industry wherein clients who apply for insurance undergo assessments to ensure that the coverage provided is equitable and appropriate in terms of benefits and affordability. The assessments paint a picture of the risk that the insurer is willing to take that the client poses and establish the appropriate insurance premiums for such risk. Underwriting involves conducting research and assessing the degree of risk that each applicant carries. These assessments focus on a policyholder’s health and related factors, in terms of health insurance, a driver’s age and safety record in terms of car insurance, or the geographical location and security of a home in terms of home insurance. The assessments aim to price insurance premiums appropriately while spreading the potential risk among as many people as possible.
If the risk is deemed too high, an underwriter may refuse coverage. If the insurance application is approved, the underwriters set premiums, and the coverage amounts based on the outcome of the risk assessment. Risk is the underlying factor in all underwriting. In the case of insurance, the risk may involve the likelihood that a prospective insured might file a claim, or that too many policyholders will file claims at the same time. The underwriting process is crucial for clients to understand, as the ability to submit successful claims on their benefits will be influenced by accurate risk disclosures made at the beginning of the policyholder’s application.
The use of AI in the underwriting process has enhanced fraud detection systems by enabling the identification of irregular patterns, mitigating the effects of human error, and reducing subjective biases. The integration of connected devices, telematics, and predictive analytics leads to a more profound understanding of market trends and customer behaviour, which in turn expedites the resolution of issues and improves customer satisfaction by providing a personalised touch over a ‘one-size-fits-all’ approach which no longer attracts modern consumers. This enables insurers to offer tailored policies and pricing by analysing customer data, leading to more personalised and relevant insurance solutions and enhancing risk management by using predictive analytics in assisting insurers to better assess risks, ensuring more reliable coverage for customers and the insurer making informed decisions by providing deeper insights into customer behaviour and market trends.
AI can handle repetitive tasks like data analysis and initial risk assessment, which allows human experts to focus on complex cases that require judgment and experience as AI is not a replacement of the human function but an enhancement. Ultimately, people will set the parameters for algorithms, monitor their performance, and make final decisions on insurance matters that are not so straightforward such as risk management, claims processing, and regulatory compliance.
The impact of using AI in insurance underwriting can be seen in:
• product recommendations, which are tailored to the preferences of clients by utilising a variety of data sources such as connected devices, wearables, speech recognition, and social media, to extract customer insights;
• optimising operational efficiency by integrating AI into policy management systems, thereby improving the overall customer experience by expediting processes, reducing manual labour and enhancing accuracy; and
• image processing and cognitive computing as insurers can conduct precise assessments and examine damages without the need for manual, expensive, and time-consuming labour. This also aids in the reduction of human error and the precise determination of the final claim settlement.
AI holds immense considerable potential for the insurance industry, however, significant concerns remain about bias in underwriting decisions and the overall fairness of these automated processes and procedures. The Organization for Economic Cooperation and Development’s (“OECD”) definition of AI is helpful as it speaks to recent forms of AI like Generative AI. According to the OECD:
“An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.”1
Datasets train AI algorithms and if the data contains inherent biases, such as discriminatory historical underwriting practices against certain demographics, the AI model will have similar biases. The effect will be unfair outcomes, such as increased premiums or potential denials of coverage. What matters to consumers is a sense of purpose, user-friendliness, and transparency in AI models. Therefore, a lack of transparency can make it challenging to identify and address potential biases within the system. If the insurance industry is perceived as unfair or biased, it can erode public trust and confidence in the system.
The future of insurance underwriting, powered by AI, not only promises a seamless experience for customers by reducing subjective biases but it also offers an efficient process with less fraudulent risk exposures to the insurer. This, however, does not curb the data security concerns that are associated with the use of AI. Insurers must ensure that customer data is collected, used, and stored according to the relevant data privacy regulations, aimed at, amongst others, protecting sensitive customer data from breaches and unauthorised access. A holistic and collaborative approach in which human expertise can be complemented by the use of AI to ensure trustworthiness, transparency and fairness is required.
