chevron-down Created with Sketch Beta.

GPSolo eReport

GPSolo eReport September 2024

AI and Insurance Underwriting: Can Artificial Intelligence and Predictive Analytics Identify Risks Better Than Humans?

Todd C Scott

Summary

  • AI is a good tool for predicting risk because it can analyze vast amounts of data quickly and identify complex patterns that humans may miss.
  • Industry experts worry that AI-powered analysis will price some people out of the insurance market.
  • AI will not likely spell the end of humans in insurance underwriting but will become a tool for insurance professionals to process more data in less time with greater accuracy.
AI and Insurance Underwriting: Can Artificial Intelligence and Predictive Analytics Identify Risks Better Than Humans?
AJ_Watt via Getty Images

Jump to:

Nikhil Rathi, the head of the Financial Conduct Authority (FCA), Britain’s financial watchdog for insurance, recently issued a warning stating that innovation in artificial intelligence (AI) risks making some people “uninsurable.” The warning issued at a September 2024 industry conference in the United Kingdom highlights concerns financial regulators have for emerging AI technologies as insurers move toward AI-powered predictive analytics for assessing risk and assigning insurance premiums.

Comparing AI insurance pricing models to dynamic ticket pricing for concerts, which often leads to surges in the cost of tickets for highly sought-after shows, Rathi warned that “Just because something can be done [in pricing technology] doesn’t necessarily mean the public will accept it.” Instead, the regulator advised insurers to adopt “Safe and responsible use of AI to drive beneficial innovation.”

What concerns Rathi and others is the rapid move toward AI-enabled hyper-personalization of insurance, where live data could lower insurance costs for some consumers while others, deemed more of a risk by the data models, could be priced out of the market, essentially making them uninsurable.

AI and Risk Analysis

It’s no surprise that AI is poised to transform the insurance industry just as it has health care, finance, manufacturing, and retail around the world. Insurance is all about mitigating financial risks by examining such factors as historical data, personal or business information, location, and behavior. AI is a good tool for predicting risk because it can analyze vast amounts of data quickly and identify complex patterns that humans may miss. By processing historical data, behaviors, and external factors in real time, AI may vastly improve accuracy in risk assessments. Moreover, AI continuously learns and adapts, making predictions more precise and personalized over time, which can lead to more effective underwriting and pricing.

Despite regulatory concerns over AI-powered pricing models, Rathi agrees that insurance risk analysis is ripe for AI innovation. If done right, AI-powered insurance pricing could benefit consumers and lead to long-term growth and productivity improvements in an industry that is often slow to change.

AI and Insurance Chatbots

AI is not entirely new to insurance. Since 2010, many of the nation’s top insurers have introduced AI-powered virtual assistants to help customers apply online. However, advancements in AI have allowed insurers to increasingly use machine learning and predictive analytics to refine pricing models in real time, ultimately producing premium quotes in seconds for customers to bind a policy online. Insurance leaders with AI assistants and automated underwriting include:

  • Lemonade. Lemonade, a disruptor in the insurance industry, has revolutionized the application process using its AI-powered chatbot, “Maya.” Maya guides customers through a simple, conversational interface to gather information, such as their location, property details, and insurance needs. Based on this data, the AI instantly issues policies in as little as 90 seconds.
  • Progressive. Progressive uses AI to enhance its online application process, allowing customers to receive customized quotes instantly. The AI gathers data from multiple sources, such as public driving records, vehicle data, and customer inputs, to generate a highly accurate insurance quote in minutes.
  • Allstate. Allstate has integrated AI into its application and claims processes using the “Allstate Virtual Assistant” (AVA). AVA helps customers complete the application process online by answering questions, guiding them through the form, and providing real-time information on their policy options.
  • Liberty Mutual. Liberty Mutual uses AI to enhance its quote and policy issuance process. Customers can receive personalized insurance quotes online by entering basic information, which AI then analyzes to generate accurate rates. The system is also integrated with third-party databases to pre-fill forms and verify customer information, speeding up the process and reducing manual input errors.
  • GEICO. GEICO offers an AI-powered virtual assistant named “Kate” that guides customers through the application process. Kate can answer common questions, help users navigate the application, and provide instant quotes based on customer input. The app uses AI to predict user needs and offer recommendations based on customer data, creating a personalized and streamlined experience.

Better Than Humans?

Colin Lachance, principal of PGYA Consulting of Ottawa, Ontario, Canada, is currently serving as the Ontario Bar Association’s Innovator-in-Residence and is spearheading a comprehensive initiative aimed at assisting lawyers with effectively integrating artificial intelligence (AI) into their practice to better serve their clients. He spoke at the 2024 North American Bar Related Insurance Companies (NABRICO) conference in Calgary, Alberta, Canada, where insurers gathered to understand AI innovation in risk assessment and what it would mean for insurers to incorporate AI into their human-driven processes.

“We’re at a point where AI is converting from a rapid answering machine to a tool for reasoning,” said Lachance in a recent interview. “That opens the door for automated data analytics. There is no going back.”

When asked why AI tools are uniquely suited for insurance processes, Lachance said, “AI is very good at identifying patterns in data. More importantly, it is also very good at identifying what’s off or wrong with the pattern.”

Spotting changes in data patterns has historically been considered a task only humans could perform and is a critical component in identifying and assessing insurable risks.

AI is also getting very good at the classification and tagging of data, according to Lachance. While humans bring judgment and contextual understanding, AI excels at processing large volumes of information quickly and objectively, leading to more accurate and personalized risk predictions.

Can AI and predictive analytics identify risks better than humans in insurance? ChatGPT confidently answers the question, “Yes.”

Lachance, however, does not see AI as the end of humans in insurance underwriting and price setting. More likely, it will become a tool for insurance professionals to process more data in less time with greater accuracy. Just as IBM’s Watson Health assists doctors in producing personalized treatment plans, highly trained insurance professionals will be needed to assess AI-produced policyholder outcomes. “AI doesn’t replace the human,” says Lachance. “It empowers the human to do deeper analysis.”

    Author