Stringent cost management is the key to insurers’ profitability. However, more recently they have been challenged by factors ranging from fast changing customer preferences, new regulations and operational costs, such as claims, administrative expenses, and personnel. In this article, we focus on how AI can reduce costs in three key use cases – claims processing, fraud detection and prevention, and risk assessment and underwriting.
AI as a technology has become the latest innovation to captivate businesses with its expansive potential. From its inception in the latter half of the 20th century to advanced Machine Learning (ML) and Deep Learning (DL) technologies, the journey has been thrilling and inspiring in equal measure. During 2023, disclosed GenAI-related equity funding skyrocketed to USD21.8bn in 2023, compared to USD4.3 billion in 2022 and USD9.9bn in 2021.1 No less than USD10bn of this total was channeled to OpenAI, developer of the ChatGPT model. This surge in new investment promises to increase productivity in every area of financial services, and insurance is no exception.
Efficient claims processing is essential for the insurance industry, impacting customer satisfaction, operational efficiency and profitability. Claims processing remains laborious, time-consuming and susceptible to errors. AI and automation have the potential to significantly increase efficiency, reduce operating costs and boost profitability. This is backed up by research, including a recent Council for the Affordable Quality Healthcare (CAQH) index report, which predicts that the automation of healthcare claims can save insurers over $11 billion annually.2 Many of these savings can be attributed to AI technology.
For example, one of the most time consuming and costly factors is the risk of errors arising from manual claims processing. By taking advantage of AI technologies such as Large Language Models (LLMs), insurers can extract relevant information from diverse sources such as claim forms, emails, and supporting documents. As well as removing the cost of human error, this accelerates the claims process and reduces demand for labor, therefore decreasing operational costs and increasing profitability.
Once the claims data has been entered, AI may also play a role in decision support systems. These systems use AI algorithms to make recommendations based on analysis of similar cases, industry standards, or policy terms. With the help of these recommendations, claims adjusters can make faster, informed decisions. AI models can also predict claim outcomes based on initial data and provide an estimate for resource allocation based on predicted claim complexity.
An example of this is how Zurich has been leveraging AI technology to predict potential outcomes from claims notifications, with a goal of pinpointing claims that may escalate, enabling early intervention and loss mitigation to benefit customers.3 By 2025, Gartner predicts that AI will enable insurers to increase customer satisfaction, reduce claims processing times by 30%, and cut the cost of claims processing by up to 40%.4
Fraudulent claims remain an enormous burden for the insurance industry—as well as customers whose premiums may rise as a result. The FBI reports that the “total cost of insurance fraud (non-health insurance related) is estimated to be more than $40 billion per year… costing the average US family between $400 and $700 per year in the form of increased premiums.”5 According to Verisk, the rate of auto insurance application fraud has gone up by 18% in 2023 compared to 2019.6
As a result, numerous insurers have implemented automated tools, including algorithms, to analyze large datasets and detect potential patterns of suspicious activity. In a study published by Novarica in 2023, 35% of large property and casualty (P&C) insurers and 24% of mid-size P&C insurers are expanding their data science efforts in order to create data sets that support anti-fraud AI tools.7
In this scenario, AI establishes a baseline of typical human behavior and screens claim in real time as they are submitted. This includes detecting collusion between claimants, service providers, and other parties involved in a claim by identifying common attributes such as phone numbers or addresses. The system flags suspicious claims for prompt investigation so that insurance companies can prevent fraudulent claims from being processed instead of taking corrective measures after the fact. As a result, insurers can process claims faster and with greater accuracy, thus reducing the cost of personnel.
Some large insurance carriers have implemented AI technology to augment existing algorithms and operations. GEICO, one of the largest auto insurers in the US, uses AI to “create better visibility into suspicious activity, improve assessment accuracy, and reduce fraudulent payouts.”8
By analyzing historical data, AI can help shape the future direction of the business. AI has the ability to analyze and predict where and when fraudulent behavior may occur in the future, allowing insurers to devise innovative strategies to help maintain profit margins.
Due to the capabilities of AI being able to analyze data sets that are too large for humans to process, it has enormous potential for risk assessment and underwriting. Manual processes, such as identifying gaps in information and obtaining approval from senior team members, also delay underwriting execution in certain lines of business, such as health insurance. Unique factors such as customer demographics, geographic location, and external risks often contribute to the complexity of assessing a prospective customer. It's essential to account for federal regulations and ethical guidelines, even when using AI to analyze data, to prevent the risk of biased outcomes.
Underwriters can use AI technology to accelerate quotations based on more accurate customer risk profiles. Leveraging Gen AI and its NLP capabilities, employees can extract information stored in different formats, such as medical reports, loss runs, and contracts. This boosts the value delivered by underwriters, broadens their business knowledge assessment, and supports well-informed decisions. In addition, Gen AI empowers underwriters to communicate and justify underwriting decisions, making it easier to manage risks faster and more effectively.
Other areas of risk assessment and underwriting where AI plays a role include:
All this enables insurers to offer customers customized and affordable policies and coverage by analyzing extensive historical data and eliminating human bias and errors.
References
1 https://www.cbinsights.com/research/generative-ai-funding-top-startups-investors-2023/
2 https://www.hyperscience.com/blog/benefits-of-automation-in-healthcare-and-insurance/
3 https://www.zurich.com/commercial-insurance/sustainability-and-insights/commercial-insurance-risk-insights/how-accurate-data-and-ai-can-transform-claims-and-help-customers-build-resilience
4 https://experionglobal.com/insurance-claims-automation/
5 https://www.fbi.gov/stats-services/publications/insurance-fraud
6 https://www.verisk.com/blog/auto-application-fraud-up-18-new-approaches-to-help-mitigate-risk/
7 https://iireporter.com/insurers-are-prioritizing-data-science-more-in-2019/
8 https://www.businesswire.com/news/home/20220928005139/en/GEICO-Extends-Relationship-With-CCC-to-Include-Smart-Digital-Fraud-Detection/
