Revolutionising the Insurance Claims Process with Generative AI
By Graham Webber, BSG Senior Consultant
Generative AI (GenAI), has been at the forefront of business and technology news over the past year. This series of articles explores how GenAI can revolutionise value chains across various industries. The focus of today's discussion is the role of GenAI in the insurance industry's claims process.
Before delving into the specifics, it is helpful to understand the broader landscape. The recent surge in the popularity of Large Language Models (LLM), the latest advancement in GenAI, has left businesses striving to comprehend the potential implications for their respective industries. Executives are simultaneously eager to explore the capabilities of GenAI and apprehensive about the appropriate deployment and commercialisation of this technology. Depending on the perspective, GenAI could either be the solution to the climate crisis and a catalyst for world peace or a threat to human freedom and the harbinger of global catastrophe. As is often the case, the reality likely lies somewhere in between. This article sheds light on the promising future of GenAI while also addressing the associated risks and mitigation strategies.
According to numerous accounts, we have already surpassed the peak of inflated expectations in Gartner's renowned Hype Cycle. As we rapidly descend into the trough of disillusionment, fear not, for we bridge the chasm and guide you onto the Slope of Enlightenment.
LLMs, a subset of AI that focuses on language and communication tasks, have been making significant noise in the industry. The launch of ChatGPT marked a turning point, shifting LLMs from a novel concept to widespread acceptance. It is crucial to remember that while this represents a significant shift for humanity, it is not a substantial technological leap forward. For instance, consider a similar improvement in a medical diagnosis AI. Initial headlines might proclaim the obsolescence of doctors, but ultimately, doctors would be the primary users of the tool, leading to improved diagnoses and treatments, . The subset of LLMs, Chat Bots, differ significantly because they are valuable to a broad user base. Therefore, their value may be inflated by the number of users, not solely by the quality of their responses.
The future state of insurance claims
Let us turn our attention to the claims process in the insurance industry. This is not a narrative about AI replacing humans, but rather about freeing up humans to allow them to ‘be more human’. GenAI can assist insurers to settle claims more quickly, accurately, and cost-effectively, while also aiding in the upskilling of their claims professionals. Claims management is not merely about processing paperwork; it involves safeguarding policyholders' interests and maintaining a trusting relationship.
An insurance claim is typically preceded by an unfortunate event. Processing a recent loss can be stressful, and dealing with a slow, inefficient claims process only adds to the burden. GenAI is set to significantly enhance claim processing efficiency by assisting at all the major steps in the claims journey.
Let us examine the claims journey and see how LLMs will impact each of the personas involved:
DISTRIBUTION
Current State
Broker: My role is to sell policies. The complexities of other options may not always be my focus if I can successfully sell one product.
Customer: The overwhelming nature of navigating through various insurance choices, ladened with industry jargon, makes comparing offerings challenging. I rely on the broker’s translation if I am not fluent in English.
Future State
The broker is empowered with an AI assistant that comprehends all our offerings and policies. This assistant can translate to and from all of South Africa's official languages, ensuring everyone involved fully understands the offerings.
Broker: My role is to sell policies. With the broker AI assistant, I can find the best option for my customer, making selling easier.
Customer: My advisor can explain all documentation in my native language and even analyse a competitor’s offer for an informed comparison.
CLAIM VALIDATION
Current State
Broker: I advise the customer to submit a claim regardless of their cover. If declined, I attribute it to the insurance company’s fault.
Customer: I am uncertain if I am covered for an incident as I do not fully understand what I was sold.
Future State
Our broker AI assistant comprehends all my clients’ policies and can answer any question on them.
Broker: I can easily answer any questions my clients have regarding their policies.
Customer: I can quickly determine if I am covered for a specific incident and validate my eligibility to claim.
CLAIM SUBMISSION
Current State
Broker: If the claim is valid, obtaining all the required supporting documentation often involves back-and-forth communication.
Customer: I am unsure if I am covered for an incident, so I may compile all the required documentation for no reason. The back-and-forth communication for valid claims creates frustration and delays payment.
Future State
Our broker AI assistant can read the claim submission form and the supporting documents, ensuring all relevant documentation is submitted, correctly filled out and legible.
Broker: When assisting my customer with a claim submission, we immediately confirm that all the relevant documentation has been submitted.
Customer: The claim submission is straightforward, and I am confident that all required information is available for quick processing.
CLAIM ASSESSMENT
Current State
Triage Agent: I manually check that a claim has all the documentation, then send it to the appropriate department.
Assessor: If additional information is required, we lose a few days. Sometimes, we must restart the claims process when more information is requested.
Future State
We automate the trivial parts of the triage process with a triage automation AI. Our assessor is empowered with an assessor assistant AI that summarises the claim and documentation.
Triage Agent: For simple cases, I verify the triage automation AI’s decision and send it to the assessor in seconds.
Assessor: Claims almost always come through with all required documentation the first time. The assessor assistant AI summarises the details and, in simple cases, gives a recommended assessment. I ask the assessor assistant AI specific questions about the claim so that I can make the right decision. The AI learns the questions that I ask it and adds them to the summary for similar queries in future.
DECISION LETTER
Current State
Assessor: We use a generic template that we complete. I copy and paste sections of the contract and documentation to justify the outcome.
Customer: The decision letter feels cold and removed. The language used to justify the decision is difficult to understand.
Future State
Our assessor is empowered with a communication specialist AI that can coach the assessor to be empathetic and translate the letter into the customer’s language.
Assessor: The communication specialist AI drafts a letter for me, prompts me to be more empathetic and personal, and assists in simplifying the language in the justification.
Customer: I can fully understand the decision and feel that my insurance company empathises with me.
GenAI also helps with behind-the-scenes processes, by creating richer, more structured data from claims documents, providing better analytics on trends, pricing, and products.
More broadly, GenAI and AI can help detect fraud by analysing the language and grammar used in the claim. They can help to upskill junior assessors, by assessing the questions more senior assessors are asking about claims, then prompting the junior assessors to consider those questions in similar scenarios.
Risk
GenAI, while a powerful tool, is not without its risks, necessitating careful consideration and robust governance. The importance of data security in preserving customer trust and complying with regulations such as the Protection of Personal Information Act (POPIA) cannot be overstated. Given the uncertain regulatory landscape for AI, staying abreast of evolving regulations is essential. While GenAI can help mitigate human bias, it is not inherently unbiased, and its outputs must be scrutinised to ensure fairness. Like any technology, GenAI is not infallible, and organisations must equip their employees to detect and prevent the acceptance of inadvertent misinformation. This is especially vital in trust-dependent industries. The ‘black box’ nature of AI, which often conceals the reasoning behind its decisions, presents an additional challenge. Therefore, organisations must strive for transparency and comprehensibility in AI decision-making processes.
The goal? Positive change
The need for GenAI in business modernisation is paramount, yet it requires a profound understanding of your business. BSG has a extensive experience in collaborating with insurers, aligning data strategies with business objectives, and transforming data into a competitive edge.
Our comprehensive approach to modernisation integrates people, processes, and technology - enhancing operational efficiency and revolutionising workflows through advanced AI and automation. Recognising the enduring presence of GenAI, we have made substantial investments to become experts not only in the technology, but also in its business impact.
Your business requires more than just GenAI; it needs enhanced efficiency, productivity, and accuracy to boost profitability. GenAI is a tool that facilitates these improvements, turning what was once science fiction into reality.
At BSG, we believe in being a proactive force for positive change. Technology can enable us to drive forward positive change, and LLMs represent significant movements in that technology.
A proactive force for positive change
As a leading homegrown South African consulting and technology company, we provide our clients peace of mind, as we help them navigate the complexities of digital strategy formulation all the way through to execution.
Get in touch with us to discuss how we can support your organisation to revolutionise your insurance claims process with generative AI.
