Generative AI: A Comprehensive Guide for Insurance Executives
We stand at a critical juncture in the insurance industry, where emerging technologies are reshaping our business landscape. Among these, Generative AI emerges as the most promising technology to revolutionize risk assessment, customer service, and operational efficiency.
Preface
Dear Insurance Industry Leader,
We stand at a critical juncture in the insurance industry, where emerging technologies are reshaping our business landscape. Among these, Generative AI emerges as the most promising one to revolutionize risk assessment, customer service, and operational efficiency.
Before we dive deeper, I thought I'd give generative AI a taste of its own medicine. I asked it to create a word cloud of the key topics we'll be covering. Let's see if it can predict its own future in insurance!
As you can see, our AI friend has a pretty good grasp of what's important in this guide. From 'risk assessment' to 'customer' focus, 'data' processing to 'underwriting' - it's all there.
In recent years, we've witnessed numerous technological hypes—NFTs, Web3, cryptocurrencies—many of which lacked substantial backing or fell victim to exaggeration and misuse. When overhyped projects fail, they often diminish the value of the underlying technology. This may even mislead industry veterans' perception of new technologies.
It's against this backdrop that I've composed this comprehensive and practical guide for insurance professionals and executives. My goal is to demonstrate concrete ways to leverage Generative AI, whether through in-house development or by tapping into external expertise.
Given the novelty and sensitivity of this topic, I've included practical examples from my technology company, Allegory. This transparency serves a dual purpose: it provides you with real-world data points and underlines my commitment to sharing actionable knowledge.
This guide is designed to equip you with the insights needed to navigate the AI revolution in insurance. We'll cover the fundamentals of Generative AI, explore its practical applications in our industry, and outline essential strategies for leading your organization into this new era.
Consider this your roadmap to harnessing the power of machine learning. It's a guide to not just adapting to the changing technological landscape, but thriving within it.
Welcome to the future of insurance. Let's explore it together.
Onur Gungor
Actuary & CEO
Understanding Generative AI
Generative AI represents a paradigm shift in artificial intelligence.
These new statistical models are capable of creating new content across various mediums – text, audio, code, images, and videos. Think of it as a digital Renaissance artist, proficient in multiple forms of creative expression, much like Leonardo da Vinci, but version 10.
These AI models, such as OpenAI's ChatGPT and Anthropic's Claude, are built on transformer architecture and trained on massive datasets. They learn patterns and relationships within the data, enabling them to generate new content that follows these learned patterns in a unique way.
What's Transformer Architecture?
To put it simply, a transformer is an advanced neural network architecture that excels at processing sequential data, allowing it to understand context and relationships in text at a remarkable scale. Its power comes from its ability to process vast amounts of information in parallel, rather than sequentially like earlier models.
The scale of this training is staggering – for instance, GPT-3 was trained on approximately 45 terabytes of text data, equivalent to about 174 million average-sized books.
For the insurance industry, generative AI opens up new possibilities in risk assessment, policy creation, customer service, and claims processing.
Imagine an AI system that can comprehend and manage complex insurance claims, such as illustrating a disability annuity plan resulting from a car accident where the optional accident benefit of the first party's insurance company is responsible.
This is now possible with purpose-built transformers for generative AI.
Capabilities & Limitations
Generative AI demonstrates impressive capabilities in natural language processing, image and video generation, code generation, and data analysis. However, it's crucial to understand its limitations.
These models may generate false statements, perpetuate biases present in their training data, and require significant computational resources.
In the insurance context, while AI can revolutionize many aspects of our operations, it's essential to maintain human oversight (or human-in-the-loop), especially in critical decision-making processes.
Further, regulatory bodies will put more scrutiny on the use of such general-purpose technologies (e.g., ChatGPT). That's why I recommend insurers to successfully conclude their efforts in digitizing their operations. This means developing the ability to store any information known to the carrier in a digitized format that has been moderated and categorized structurally.
Implementing this digital transformation will have a profound impact on your organizations, propelling your success for more than a decade and making your organization more resilient to emerging risks for the future.
The key is to leverage AI's strengths while mitigating its weaknesses through careful implementation, continuous monitoring, and expert support.
Impact on the Insurance Industry
The impact of generative AI on the Property and Casualty (P&C) insurance industry is profound and far-reaching.
McKinsey's 2023 report suggests that three-quarters of work activities consuming employees' time today could potentially be automated with generative AI. Given the persistent expense ratios over the past two decades, this technology can now make a significant dent in operational costs and efficiency.
To put this into perspective, a 5-percentage-point reduction in the US P&C industry's expense ratio through the use of generative AI could result in an astounding $50 billion annual savings. This level of cost reduction has the potential to revolutionize the industry's profitability and competitiveness.
Here's how AI is transforming various aspects of P&C insurance:
Risk Assessment, Underwriting, and Pricing
Generative AI is revolutionizing risk assessment through real-time risk profiling and advanced catastrophe modeling. It enables more accurate and personalized underwriting decisions, leading to better risk selection and pricing.
For instance, machine learning models can analyze data from IoT devices and satellite imagery to provide up-to-the-minute risk assessments for properties and vehicles. What's more, they can do so much faster and more affordably than current methods, significantly improving the speed and accuracy of risk evaluation.
At Allegory, we've pushed the boundaries of this technology. In one of our projects, we developed a feedback system for usage-based insurance telematics. This system goes beyond traditional telematics by incorporating generative AI:
- We feed the AI system with week-by-week driving behavior data
- The AI analyzes this data to understand patterns and risk factors
- Based on this analysis, we generate highly personalized risk mitigation advice for drivers and their family members
- This advice is then used to inform underwriting decisions and pricing in real-time
This real-time, AI-driven approach to risk assessment and underwriting represents a significant leap forward in the insurance industry. It allows for more dynamic pricing models, better risk management, and ultimately, a more personalized and fair insurance experience for customers.
Product Development, Distribution, and Sales
AI is revolutionizing the creation of highly personalized insurance products and customized policy wordings.
These advanced systems analyze customer data with a focus on intent, allowing for proactive suggestions of additional coverages or policy adjustments. This leads to more comprehensive and tailored coverage, enhancing both customer satisfaction and revenue potential.
The return on investment (ROI) for implementing this technology is undeniably positive, with a payback period of less than one year. This makes it an attractive option for insurers looking to innovate and gain a competitive edge.
To leverage this technology, your organization can start by building an agile team (or a "tribe" as seen in some insurers) dedicated to collecting and classifying current and historical policy documents.
By combining these with coverage information and quotation data, a dedicated generative AI algorithm can be trained specifically on your own data. This approach ensures that the AI is tailored to your specific products, market, and customer base.
The impact of this technology is far-reaching across all distribution channels.
Whether you distribute through digital channels, brokers, or captives, this technology empowers the end user in composing 100% tailored products, in any branch, at the fingertips of your partners.
It's particularly beneficial for converting and retaining high-net-worth individuals and securing high-value commercial customers.
At Allegory, we've taken this concept further with our risk-based hyper-targeting product. This innovative solution provides sales staff, technical and on-site underwriters, and Chief of Revenue with a powerful tool to visualize their mass experience in their field and act immediately on insights gained.
This visual representation enables the identification of high-potential areas for specific products, tailoring of marketing and sales strategies to local risk profiles, and more efficient allocation of resources in sales and underwriting efforts.
By leveraging AI in this way, insurers can significantly enhance their distribution and sales strategies, leading to increased market share and improved customer retention.
The technology provides an avenue to distill and visualize mass experience in the field and act immediately, giving your team a significant advantage in a competitive market.
Claims Processing, Fraud, and Accounting
Generative AI is revolutionizing the claims journey through AI-powered damage assessment, predictive triaging, and virtual claims adjusting. This transformation leads to faster resolution and improved customer satisfaction.
Natural language processing now enables automatic claim initiation from customer calls or messages. Simultaneously, AI-powered image analysis can estimate repair costs and detect potential fraud. Generative AI's ability to process these tasks concurrently marks a significant leap forward in claims handling efficiency.
Fraud detection and prevention stand out as a prominent use case for generative AI. Drawing from my experience as a buyer of such technologies and as a member of The Association of Certified Fraud Examiners (ACFE), I've observed that generative AI excels at what previous technologies attempted: understanding complex networks and revealing hidden relationships.
By simply processing your historical claims and related correspondences through generative AI, you can:
- Uncover sophisticated fraud rings
- Assist internal auditors in proactive response
- Enhance overall risk management
In response to this, at Allegory, we've developed an innovative solution that allows underwriters to identify discrepancies between underwritten mileage and actual mileage. This tool exemplifies how AI can bridge gaps in information, improve underwriting accuracy, and significantly reduce premium leakages.
The power of generative AI in claims and fraud detection lies in its ability to analyze vast amounts of data simultaneously, identifying patterns and anomalies that human analysts might miss. This capability not only speeds up processes but also significantly improves accuracy in fraud detection and claims handling.
Information Technology (IT) and Technical Debts
While many insurers remain cautious about adopting machine learning applications in their customer-facing processes, generative AI offers an immediate benefit for your company's IT department.
As mentioned earlier, this technology demonstrates impressive capabilities in natural language processing, as well as in code generation and data analysis. These capabilities can be particularly valuable in addressing technical debt.
Many banks and insurance companies still rely on older programming languages due to their reliability, performance, and the massive legacy systems built with them. Generative AI can potentially help:
- Identify bugs in legacy code
- Upgrade core components to make them less error-prone
- Refactor mainframe applications in-house
Below is a concise table outlining the usage of each language commonly used by insurance companies and how generative AI can assist in improving efficiency or refactoring:
| Programming Language | Usage in Insurance | Generative AI Benefits |
|---|---|---|
| COBOL (Common Business-Oriented Language) | Transaction and batch processing on mainframe systems | Automates code optimization and documentation; assists in identifying and fixing bugs |
| RPG (Report Program Generator) | Business applications on IBM iSeries (AS/400) | Streamlines code modernization; facilitates migration to modern platforms |
| PL/I (Programming Language One) | Commercial and scientific applications on mainframes | Enhances code readability; accelerates refactoring processes |
| Assembler Language (Assembly) | System-level programming in mainframes | Converts low-level code to higher-level languages; detects inefficiencies |
| FORTRAN (Formula Translation) | Financial modeling and computations | Translates legacy code to modern languages; improves computational efficiency |
| JCL (Job Control Language) | Job scheduling and execution on IBM mainframes | Automates script generation; optimizes job workflows and scheduling |
This application of generative AI can significantly reduce the time and resources required for maintaining and upgrading legacy systems. It offers a pathway to modernization without the risks associated with complete system overhauls, potentially saving insurers millions in IT costs and reducing operational risks.
Preparing Your Business for the AI Revolution
As an actuary and entrepreneur deeply immersed in the insurance industry, I strongly advise companies to initiate the implementation of generative AI technology. Your actuaries and data scientists are uniquely positioned to spearhead this transformation within your organization, leveraging their analytical skills and industry knowledge.
Based on my experience working with executives across the industry, I recommend the following seven key steps for successful AI integration:
1. Assess AI Readiness
Conduct a thorough evaluation of your current data infrastructure and identify potential use cases within your organization. This assessment should encompass data quality, accessibility, and the technological capabilities of your existing systems.
2. Develop an AI Strategy
Craft a comprehensive AI strategy that aligns with your business objectives and outlines a clear roadmap for implementation. This strategy should consider both short-term wins and long-term transformational goals.
3. Invest in Data Infrastructure
Prioritize investments in data quality and infrastructure. This includes implementing robust data governance policies and adopting scalable cloud computing solutions to support AI operations.
4. Build AI Talent
Cultivate AI expertise through a combination of upskilling programs for existing staff, strategic hiring of specialized talent, and partnerships with AI consultancies or academic institutions. This multi-faceted approach ensures a well-rounded AI capability within your organization.
5. Initiate Pilot Projects
Begin with carefully selected pilot projects to gain practical experience and develop a scaling framework for successful initiatives. These pilots should be designed to deliver measurable value while providing learning opportunities.
6. Address Ethical Concerns
Proactively tackle ethical considerations by establishing a cross-functional AI ethics committee and implementing regular AI audits and impact assessments. This approach safeguards your reputation and ensures responsible AI deployment.
7. Foster Innovation Culture
Cultivate a culture of innovation by establishing an 'AI Innovation Lab' and organizing cross-functional AI hackathons or innovation challenges. This environment encourages creative problem-solving and collaborative AI development.
8. Leverage Regulatory Compliance as an Asset
A unique advantage for insurers in adopting generative AI lies in the vast repository of previously approved filings, documents, and procedures. By using this pre-approved data as input for AI models, companies can potentially streamline regulatory compliance for AI-generated outputs. This "approved filings in, approved filings out" approach could significantly accelerate AI adoption while maintaining regulatory alignment.
Conclusion: Leaders and Followers
Generative AI represents a paradigm shift in technological advancement, with profound implications for the insurance industry. Its potential to enhance risk assessment, personalize policies, streamline operations, and improve customer service is unparalleled in recent history.
The future landscape of insurance will be defined by those who can effectively harness the power of generative AI while adeptly navigating its challenges. By taking proactive steps to prepare your organization, you position your company at the vanguard of this AI revolution.
It's crucial to remember that while intelligence is replicable, creativity lies in the randomness. Your unique vision, combined with AI's capabilities, will be the driving force behind true innovation in the insurance industry.
As you navigate this transformative journey, maintain your curiosity, stay informed of rapid developments, and above all, preserve the human element in your operations.
The future of insurance isn't solely about AI – it's about how we leverage AI to better serve and protect our customers, enhancing the very essence of what insurance represents: security, trust, and peace of mind.
Welcome to the future of insurance. The time to lead is now.