AI’s Role in Transforming the P&C Insurance Landscape

Advances in artificial intelligence, notably generative AI technologies including Large Language Models (LLMs) such as OpenAI’s ChatGPT, promise a further tectonic shift in the insurance landscape.

(Image credit: Igor Omilaev/Unsplash.)

For the past few years, P&C insurers have been primarily focused on modernizing core systems, including those pertaining to policy administration, claims, billing, and broker management. Innovations such as the Internet of Things (IoT), sensors, wearable devices, gamified applications, blockchain, driverless cars already present opportunities for insurance carriers and other industry players to reshape the design and delivery of products and customer experiences.

Now GenAI is set to further accelerate this revolution across a range of insurance-related activities and functions, from distribution, rate making and underwriting through to claims adjusting, subrogation, customer service, and fraud detection.

Claims Adjusting

AI can help adjusters and insured to speed up the claims process. Machine learning (ML) can be used to create predictive models that allow claims adjusters to make faster, better, and more informed decisions at the first notice of loss (FNOL), for example in the case of an automobile collision. ML can offer immediate insight into the repairability of a car based on predictive analytics, so an adjuster does not have to wait days for an appraiser to physically inspect the vehicle or for a repair shop to tear it down to assess the damage.

If a total loss, the insurer can get ahead working with the insured on settlement, and the insured can start looking for a replacement sooner. If the vehicle can be repaired, AI can expedite the process by locating the closest mechanic or body shop with the most relevant expertise for the work required.

For accident benefits and bodily injury claims, AI can help assess injuries resulting from automobile or other accidents, with GenAI capable of ingesting and summarizing large volumes of medical data in seconds. Documents associated with claims, whether collision reports, treatment reports or litigation documents, then tie this to historical data on what has influenced claim outcomes. Spending less time aggregate data means more time to be spent on the litigation. It can lead litigators to win court cases, and faster return to work for the injured.

GenAI also has a part to play in training newly recruited claims adjusters. For example, AI can simulate a conversation with an angry or dissatisfied customer and then provide invaluable tips and guidance based on the adjuster’s performance through the use natural language processing (NLP), natural language understanding (NLU), sentiment analysis and tone analyzer to detect stress levels to mitigate fraudulent claims.

Rate Making and Underwriting

Having traditionally relied on more conventional tools and techniques for actuarial rate making, insurers now have the opportunity to use AI in rate setting and underwriting to evaluate risk on an increasingly granular basis. This can lead to greater accuracy and fairness in policy pricing and hyper personalized risk assessments.

AI can help in building advanced pricing models, with AI algorithms sifting through insurers’ extensive databases and identifying new risk factors often overlooked by traditional methods. Insurers not only have access to claims data about the insured but also can leverage industry wide data.

Insurers have to deal with both structured and unstructured data. Structured data is highly organized and easily decipherable by machine learning algorithms, whereas unstructured data cannot be easily processed and analyzed via conventional data tools. Examples of unstructured data might include emails, voice recordings, images and videos.

Forward-looking insurers are building data lakes from which raw data can be extracted for processing by machine learning algorithms. They are also drawing upon other data sources, such as social media, IoT devices and wearable technologies, to gain more detailed, real-time actionable customer insights.

AI-powered telematics in the auto insurance space is another example, analyzing data captured via devices in vehicles relating to speed, braking habits, and driving times. Insurers can use this to offer usage-based policies, with premiums tailored to the insured’s driving behavior, with a view to encouraging and rewarding responsible driving.

The availability of ever more complex datasets and advances in predictive analytical tools opens the potential for insurers to cover risks that were previously seen as uninsurable.

Detecting Fraud

Insurers are using AI and Machine Learning algorithms to automate claims processing. The system reviews data from various sources, including images, videos, and audio recordings, to assess claims more accurately. Insurers are also able to detect potential fraudulent claims by identifying patterns that indicate fraudulent behavior—investigating and establishing such connections are currently extremely time consuming and resource intensive.

In addition to sediment analysis and tone analyzer mentioned earlier, AI can also be applied to verifying the authenticity of images via the forensic analysis of image data to determine when and where a photo was taken and if it is a downloaded image from the Internet.

Most programming in current fraud detection applications is rule based. The program is instructed to recognize a particular kind of evidence as suspicious and knows to flag those cases to investigators. Rules-based engines are relatively easier to develop, but there are challenges around adding new rules or even knowing which rules to include the first place.

Machine learning allows insurers to detect fraudulent patterns by analyzing the patterns rather than just hard coding rules for the engine to follow. GenAI can help investigators by handling tedious tasks such as reviewing hundreds of pages of documents, freeing them up to spend more time on arbitration and litigation.

Data Privacy and Bias

Public Large Language Models such as Open AI ChatGPT, Microsoft Co-Pilot, Google Gemini and Anthropic Claude typically encode large volumes of data drawn from across the internet, but LLMs can also be fine-tuned—and the quality of their outputs enhanced—by training them using much more carefully curated data.

The protection of personal data collected, processed, and stored by public LLMs is another consideration given the potential risk of data leaks and improper access. AI models need to incorporate guardrails and safeguards to limit unnecessary data exposure and make sure sensitive data is not passed to public LLMs.

Bias in AI models is also a concern, it can skew results and also lead to discrimination. To avoid such issues, insurers must be able to explain why they have used certain data inputs and how their AI drew its conclusions.

The European Union recently adopted its own AI Act to address the rapid advancement and integration of AI technologies, while in the U.S. President Biden last year issued an executive order aimed at strengthening AI safety and security.

AI, GenAI: What’s Ahead for the Insurance Industry?

 

 

 

Behzad Salehoun and Serena Chan //

Behzad Salehoun is a partner and the Canadian Insurance practice leader at Capco, a global technology and management consulting company focused on innovation in financial services. With 20 years of experience in Insurance and Financial services he is a trusted advisor and a transformation leader, focused on helping clients achieve their strategic goals and deliver value to their customers. He is passionate about insurance innovation and Insurtech and has successfully delivered industry-leading digital and data solutions for top Canadian and global clients.

Serena Chan is a Partner and Head of Innovation and Design Practice with Capco Canada with more than 25 years of consulting experience including PwC Consulting, BearingPoint, IBM Consulting and Capco. She is a Chartered Insurance Professional (CIP) and a 4-time IBM Redbook author. Serena has led portfolios of work with clients who have designed new and differentiated legendary customer experiences, developed and implemented award-winning solutions, and adopted AI technologies for customer experience and operational efficiency.

Leave a Comment

(required)