AI in Insurance: 4 Critical Considerations

AI in Insurance: 4 Critical Considerations

Artificial intelligence (AI) is transforming the world, and the insurance sector is no exception. There is already a wide range of AI-powered solutions for functions from claims management and customer service to fraud reduction, pricing decisions, and underwriting efficiency. 

Technology experts say AI is here to stay, so the question facing insurance leaders today is when and how to adopt AI-driven technologies. Leaders need to address some key considerations as they answer that question: 

1 – What are the limitations of AI-based tools? 

According to Anthony Habayeb, CEO and founder of Monitaur, an AI governance and machine learning company, a major limitation of AI is the programming itself. AI data models used to analyze massive amounts of data and draw decision-useful conclusions are programmed to find correlations. They are, however, incapable of finding correlations or taking context into consideration when drawing those conclusions. This can lead to discriminatory outcomes. 

For example, if a life insurance company applies a model that uses body mass index (BMI) as a factor in estimating life expectancy, the model may find a correlation between high BMI and mortality rates, driving price increases for those with higher BMIs. The context not taken into account is that Black Americans tend to have a high BMI. As a result, the higher prices would have an outsized and discriminatory effect on the Black population. 

In addition, the types of AI used in the insurance industry are limited in their ability to access information. Therefore, models may draw incorrect conclusions if they are not regularly updated with current information. 

Habayeb concludes that it is imperative for organizations to build a system of checks and balances for the AI, including “planning, approval, auditing, and continuous monitoring” to compensate for the limitations of AI programming and avoid negative outcomes. 

2 – What risks could AI-based tools introduce to an organization?  

New tools can impact a company in unexpected ways. The same is true of AI-based insurance solutions. 

Within an industry as heavily regulated as insurance, those impacts can pose significant risks. Habayeb identifies 5 types of risk to AI adoption: 

  • Reputational Risk: This focuses on public perception and customer experience. Firms must ask whether the technologies may negatively impact public image and customer loyalty and take steps to mitigate. 
  • Legal Risk: Since AI and ML are relatively new, legal and regulatory requirements are still developing. Companies using AI are currently operating in a gray space and should self-monitor to ensure the ethical use of AI. 
  • Strategic & Financial Risk: As AI takes on an enterprise-level strategic role, the risk of large-scale mistakes with significant ramifications increases. Companies should take steps to prevent AI from impacting revenue streams and ensure that the assumptions used by their AI models are as accurate as possible. 
  • Operational Risk: It takes time to identify limitations, drawbacks, and security risks in new technologies, so organizations should take their time when testing AI solutions to ensure that prospective tools meet all operational requirements and don’t pose risks to critical operations.  
  • Regulatory & Compliance Risk: The regulatory environment for AI use in insurance is still evolving, causing many unknowns. Firms should keep tabs on emerging requirements from regulatory bodies and be prepared to adapt quickly to achieve compliance. 

These 5 types of risk are significant, but not necessarily insurmountable. 

3 – How much risk is acceptable? 

With many unknowns and the potential for wide-reaching risks, firms may question how much risk they should accept when adopting AI solutions. In Deloitte’s recent article, “Risk and compliance implications of AI in the Insurance Industry,” the authors identify three approaches to addressing the risks posed by AI.  

The first approach is risk averse. Organizations hold off on AI adoption, waiting for stricter and clearer regulations. They avoid the rocky early years of technology development and maintain a traditional approach that retains customer trust. However, they risk falling behind technologically as the industry develops and may not be able to recover the ground they lost. 

The second approach is that of the early adopter. Firms embrace technology rapidly and are known as pioneers. This mindset assumes that, to borrow Deloitte’s phrasing, “innovation will win out over risk.” However, organizations that take this approach have significant risk exposure that could cause negative impacts to the organization. 

The third approach strikes a balance. Firms use a risk-based approach to technology adoption, first implementing a risk and compliance management strategy to guide the adoption of AI solutions. They reap the benefits of technology adoption while minimizing risks. 

While each approach offers pros and cons, organizations must determine which stance aligns best with their values and executive leadership vision. 

4 – What does a successful technological transformation look like operationally? 

In addition to considerations around the limitations and risks of AI, there are practical questions regarding successful implementations.  

A report from McKinsey, titled “Insurance 2030—The impact of AI on the future of insurance,” offers insights into this issue. The first step, they say, is to learn about AI, evaluate possible solutions, and take time to assess “how to use technology to support. . . business strategy.” Then, firms should develop a strategic plan that addresses all facets of the technology transformation, including change management, data capabilities, and talent. 

The talent side presents challenges since many employees fear AI will have a negative impact on their work. Kannan Amaresh, SVP & Global Head of Insurance at Infosys, says in a Forbes article that it is vital that leaders take steps to make employees feel included in the process to alleviate fears about job loss and AI’s impact on their roles. He also observes that workforce retraining to focus on strategic rather than repetitive tasks is key to a successful organizational shift. 

Ultimately, as McKinsey’s report notes, a successful transformation “will require a conscious culture shift. . . that will rely on buy-in and leadership from the executive suite.”  

The Way Forward 

The role AI will play in the insurance industry is still evolving. It has inherent limitations and poses risks but also offers significant rewards, particularly to organizations that approach adoption and address risks and relevant workforce concerns thoughtfully.