How to Utilize AI in Insurance Underwriting
Artificial intelligence (AI) in insurance underwriting is revolutionizing carriers’ business models by enhancing risk assessment, personalizing customer offers, and streamlining processes across functional boundaries.
AI brings unique capabilities to insurers through its ability to analyze vast datasets rapidly, so those insurers can make better-informed decisions and improve their operational efficiency and profitability.
AI enables the rapid assimilation of seemingly unrelated data into a coherent view of risk on a more granular basis. This allows for the crafting of personalized and compelling offers, and the pricing of those offers to maximize competitive position and attractiveness for each prospect.
Incorporating AI and automation into underwriting workflows can also reduce the time spent by underwriters on administrative tasks and organizational maneuvering, resulting in substantial productivity gains.
In this blog post, we will look at the following:
An overview of AI in insurance underwriting, including key use cases for property and casualty (P&C) insurers;
The benefits AI in insurance underwriting brings;
How to overcome some of the barriers and issues, both real and perceived, in AI-driven underwriting; and,
What insurers should do now to take full advantage of AI-enabled underwriting.
An Overview AI in Underwriting – Four Key Use Cases
Artificial intelligence (AI) is transforming insurance underwriting by enhancing risk assessment, delivering personalized policy offerings, and streamlining a variety of processes. By leveraging AI, insurers can analyze vast datasets with unprecedented speed and accuracy, leading to more informed decisions and vastly improved customer experiences (CX).
AI in insurance has gained a significant foothold, with a majority of insurers at a minimum piloting, and with many now in production, leveraging technologies such as large language models (LLMs), machine language (ML), and now generative AI (GenAI).
This chart illustrates the adoption rates for these technologies across various insurer functions, according to Conning, a global investment management firm that tracks the insurance industry:
Use Case #1 - Enhanced Risk Assessment
AI enables insurers to process extensive amounts of structured and unstructured data, such as historical claims, customer behavior, and external factors such as competitive actions, market trends, and social media activity.
This comprehensive analysis allows for more precise risk evaluations, facilitating tailored coverage for clients, and bolstering carriers’ risk management strategies.
Use Case #2 - Personalized Policy Pricing
AI facilitates personalized policy pricing by evaluating individual risk profiles, leading to more accurate pricing models.
This approach will not only attract a broader customer base, but also complements elements of risk management. For example, AI-driven systems can calculate prices with greater precision to ensure policyholders pay rates that match their actual risk profiles.
Use Case #3 - Automation of Underwriting Processes
Integrating AI into underwriting automates routine tasks, reducing manual intervention and expediting decision-making, making for much more efficient underwriting and underwriters.
For instance, AI can analyze customer documents and risk profiles, automate application approval decisions, and recommend optimal coverage terms based on individual risk scores.
Use Case #4 – Incorporating Generative AI (GenAI)
Generative AI (GenAI) has captured the popular imagination, driven unprecedented sums of investment capital, and has already advanced human productivity in a number of areas.
In insurance, Gen AI has been applied in several key areas of underwriting:
Natural Language Processing for Document Analysis
Underwriters deal with vast amounts of paperwork, including claims documents, financial statements, and legal documents. Generative AI-powered Natural Language Processing (NLP) can assist by efficiently extracting key insights, summarizing critical information, and even drafting summary reports, thereby reducing manual effort and increasing underwriters’ efficiency.
Personalized Policy Generation
Working from the risk analysis and pricing recommendations referenced above, GenAI can then be used to generate highly-personalized policy documents more efficiently and effectively, saving time, improving profitability, and further reinforcing personalization and customer engagement.
Generating Synthetic Data for Model Training
To improve underwriting models, generative AI can create synthetic datasets that mimic real-world scenarios. This allows for training and retraining analytical models without exposing sensitive customer data, thus forwarding the cause of responsible AI by ensuring compliance with data privacy regulations and improving transparency in underwriting transactions.
Chatbots and Virtual Assistants for Underwriting Support
In customer-facing applications, GenAI-driven chatbots and virtual assistants can guide customers and brokers through the underwriting process, answer questions, and collect necessary information. These tools improve customer experience and reduce the workload on human underwriters.
A special class of GenAI-based virtual assistants known as copilots can also be brought to bear for improving the productivity of internal carrier staff, with applications such as rapid knowledge assimilation and specialized coding assistants.
Benefits of AI in Insurance Underwriting
AI has already shown significant real-world results and demonstrable ROI in the global insurance industry.
Advances in Risk Mitigation
In order to improve margins for insurers, risk mitigation is key to maintaining profitability. The use of AI in insurance underwriting can be a key contributing factor to this quest.
Reports of several quantitative improvements in this arena have borne this out. The use of AI and ML models in underwriting have delivered improvements in risk mitigation that include:
Combined Ratio improvements of 3-6 percentage points;
Loss Ratio improvements in the 2.1-4.2 percentage point range; and,
Anti-selection of portfolio improvements of 10-15%.
Increased Revenue from Improved Pricing Personalization and Attractiveness
When it comes to capturing new business and retaining customers at renewal time, studies have repeatedly shown that price is the number one consideration for both prospects and customers.
Using AI in the underwriting process can deliver a new level of pricing intelligence, enhancing carriers’ ability to price attractively and to close more business.
Quantitative successes bear this out, as insurers have shown
Additional GWP growth of 3-4%
There is also a customer satisfaction gain to be had. By streamlining the underwriting process, AI can make it faster and simpler for customers to find the policy that best fits their needs, without spending hours or days researching different options or shopping among a slew of carriers. This can result in a sustainable competitive advantage for Insurers who leverage AI in insurance underwriting.
Increased Underwriting Productivity
Historically, the interrelated functions of pricing, rating, and underwriting have not been well-integrated, resulting in inefficiencies and an increasing probability of errors, rework, and regulatory exposure.
Automating underwriting in conjunction with pricing and rating holds much promise in breaking down organizational barriers and making the entire process smoother, more efficient, less costly, and less prone to regulatory issues:
According to McKinsey, insurers who adopt AI in underwriting have achieved processing times up to 70% faster, reducing costs by as much as 30%.
By another account, artificial intelligence can drive as much as a 50% increase in underwriters’ productivity.
And in another report:
AI has decreased average underwriting decision times from 3-5 days to approximately 12.4 minutes for standard policies;
While maintaining a 99.3% accuracy rate in risk assessment; and,
For complex policies, a 31% reduction in processing time,
While demonstrating an improvement in risk assessment accuracy of 43%
Overcoming Real and Perceived Challenges of AI in Insurance Underwriting
As pointed out by Risk Management Magazine, using AI in insurance underwriting has certain pitfalls that must be avoided or overcome to ensure fairness, transparency, and regulatory compliance.
Some challenges, such as protecting data privacy and ensuring regulatory compliance, are not unique to AI-driven underwriting, and are the subject of intense scrutiny in all insurance companies.
Other challenges, if not planned for, mitigated, and monitored, can be magnified by the use of AI in insurance underwriting, and merit additional attention:
Avoiding bias in AI-generated decisions, due to the “black box” nature of AI models.
The use of AI, supplemented by human judgment and open communications, can assist in this area, by automating the assessment of true customer risk profiles and setting prices accordingly, creating a more equitable system and removing conscious or unconscious biases.
Ensuring transparency and fairness in AI-driven underwriting models, which is crucial to gaining and maintaining customer trust.
The power of AI to gather vast amounts of data from multiple sources and assess its accuracy and usefulness allows carriers to determine a customer’s true risk profile, leading to better decisions and a fairer assessment of the correct premium.
In many respects, insurers are no strangers to managing these downsides and leveraging technology to assist in their management. While the use of AI in insurance underwriting might underscore their importance, these risks are certainly not insurmountable or unsolvable for modern carriers.
Summary and Next Steps
AI is transforming insurance underwriting by improving risk assessment, personalizing pricing, and automating common underwriting tasks, leading to more streamlined, predictable, customer-centric, and profitable insurance businesses.
As you consider how to best take advantage of AI in insurance underwriting, it’s time to contact Earnix and learn more, including seeing a comprehensive demo.