In my 20 years in the insurance technology industry, I have seen significant technology advancements but nothing as exciting as artificial intelligence (AI).
Integrating AI in group insurance can produce tangible benefits for carriers, particularly by streamlining quoting and rating, optimizing resources, automating mundane tasks and making underwriting more accurate.
Deloitte's “Future of Insurance Underwriting” report finds the deployment of automation, alternative data and artificial intelligence (AI) are the top three changes insurers need to make in underwriting to stay resilient and set the stage for growth.
Despite such a clear need for urgency, the group insurance industry has been reluctant to adopt AI for many reasons, including regulatory challenges, the lack of available data to train AI algorithms and fears of technology replacing human workers.
What's Holding the Industry Back?
Transparency & Regulatory Challenges
AI systems can be a black box, where data goes in, results come out and nobody, not even the developers of the AI, knows how it came to its conclusions.
Here's the problem: Pure machine learning analyzes data in an iterative fashion to develop a model, and that process is not easily understandable.
Group insurance is a highly regulated industry, and regulations are moving toward carriers not being allowed to make underwriting decisions that affect their customers based on black-box AI.
For example, the E.U. has proposed AI regulations that mandate AI used for high-risk insurance applications be “sufficiently transparent to enable users to understand and control how the high-risk AI system produces its output.”
Additionally, without proper transparency into how AI underwriting systems come to conclusions, underwriting teams won't trust the system to make correct decisions, and carriers will leave themselves vulnerable to the risks of being deemed biased.
The New York State Department of Financial Services says they’re concerned insurers could use their algorithmic underwriting systems to discriminate against consumers illegally. New York’s insurance laws and similar laws elsewhere prohibit using race, national origin, lawful travel, mental or physical disabilities or traumatic experiences such as domestic abuse in any aspect of insurance underwriting.
Group insurers must demonstrate transparency, fairness and accuracy in their AI system's pricing to maintain customer trust and retention.
See also: AI: The Future of Group Insurance
Lack of Available High-Quality Data
AI algorithms need to use large data sets to work effectively.
In group insurance, brokers or employers often provide carriers with RFP information, historical claims data and census data that can be cleaned by AI tools and sent to AI-powered underwriting platforms for automated quoting and rating.
While this data is critical for AI underwriting solutions, employers typically manage the data related to group insurance policies, which can restrict carriers from accessing and ingesting more granular employee data (e.g., medical records, health screenings, past claims data, etc.) into their underwriting systems.
Additionally, leveraging sensitive plan member health data comes with the risk of infringing on consumer privacy and potentially leaking information to other parties.
A study by the Office of the Privacy Commissioner of Canada says 89% of Canadians are at least somewhat concerned about people using online information about them to steal their identity, including 48% who said they are highly concerned about identity theft.
The E.U.'s General Data Protection Regulation (GDPR) is one of the world's most extensive data compliance regulations for the insurance industry. It is designed to harmonize data protection laws across the E.U. Insurers that are based in the European Union or that process the personal data of E.U. citizens need to comply with the GDPR. The U.K also has a GDPR post-Brexit with similar concerns.
The GDPR places substantial restrictions on processing special categories of sensitive data such as race, religion, sexual orientation, sensitive health information, etc.
Carriers and vendors must comply with data privacy regulations, ensure the confidentiality of personal health information and be transparent to customers and regulators when using external data sources to increase premiums or deny coverage.
The Fear of AI Replacing Humans
Many insurance executives are concerned that AI systems will replace human workers. AI will result in cutbacks for some areas. For example, recent research by McKinsey suggests that 25% of the insurance industry is projected to be automated by AI and machine learning techniques by 2025. And according to a global survey by Rackspace, 62% of insurers have cut staff due to the implementation of AI technologies.
Yet, as AI in group insurance sales and underwriting matures and carriers gain access to new data sources rather than being replaced, many roles will be upskilled and retuned to accommodate new technologies and new ways of working. Many underwriting tasks that AI can automate are mundane, time-consuming tasks, such as converting raw data from RFPs into structured formats and manual data entry. AI can help skilled underwriters focus on more urgent and important work.
It is estimated that AI will increase labor productivity by about 37% by 2025 by eliminating or minimizing more manual tasks and freeing current workers to add more value.
See also: The Risks of AI and Machine Learning
Best Practices for Integrating and Maintaining AI Systems in Group Insurance
Clearly Define Long-Term Objectives Before Integration
Employee benefits insurers that have at least dipped their toes into AI technologies tend to use them to address narrow topics rather than high-value problems.
When carriers do not see sufficient returns on their AI investments, they may hesitate to dedicate enough money, time and attention to generate significant financial benefits. Short-term thinking and looking for quick wins do not give AI solutions adequate time to learn and prove their value. Instead, carriers should define one foothold problem within the value chain for an AI solution to solve or an opportunity to exploit.
Identifying the Problem:
Understanding the specific pain points that plan members and employers encounter is essential. These problems might include difficulties in navigating the insurer's website, unanswered simple queries or the inaccessibility of contact information. Identifying these issues is the first step in determining whether AI-based chatbots are the right solution.
Choosing the Right Solution:
AI chatbots are more sophisticated but also more expensive and complex to develop and maintain. In cases where simpler solutions can resolve the issues, investing in AI might not be cost-effective.
Defined Business Objectives:
Many AI projects fail because they lack well-defined business objectives. It's vital to have a solid understanding of what you want to achieve with AI-based chatbots. Is it improving customer service, reducing costs or increasing sales? Knowing the purpose of AI in your business is critical to its success.
AI implementation often requires patience. It's not a quick fix, and it may take time to fine-tune the chatbot for optimal performance. It's important for organizations to have realistic expectations and be willing to invest time and resources for AI to deliver the desired results.
Position for Success:
Insurers that start with well-defined business objectives, a clear understanding of the problems they aim to solve and a commitment to patiently see the project through are in a prime position to succeed. Success in AI implementation often hinges on a strategic, long-term approach.
In the insurance industry, AI chatbots can bring substantial benefits by improving customer service, automating routine tasks and increasing efficiency. However, success is contingent on careful planning, problem identification and aligning the technology with specific business needs and objectives.
See also: 3 Key Uses for Generative AI
Leverage New Data Sources
Without comprehensive historical and real-time data about plan members and the business, group benefits AI systems, such as an underwriting platform, can struggle to accurately produce quotes and rates that reflect the group's risk, resulting in financial losses.
As reflected in Majesco’s annual SMB customer survey report, many group L&H carriers are using new data sources for underwriting, including data from prescription drug purchases, fitness trackers and social media.
Fitness devices can track daily steps, sleeping patterns, activity levels, heart rates, calories consumed, etc. Its data-tracking capabilities and consumers' desire to share such information for incentives make fitness devices one of the most promising new data sources for group insurance underwriting.
Insurers that can effectively capture new data sources for their AI underwriting models will be able to deliver more accurate quotes, rates and personalized policies faster than the competition.
Use Synthetic and Internal Data
Synthetic data is not a new concept, but it is becoming a valuable resource for training AI systems. According to a report by Gartner, 60% of all data used in the development of AI will be synthetic rather than real by 2024.
Obtaining the right data is critical to training and maintaining robust AI solutions. However, collecting quality underwriting data from the real world has historically been complicated and time-consuming for group insurers.
Synthetic data refers to artificially generated data made by generative machine learning algorithms and statistical models. Its ability to replicate the characteristics and signals of real genomic datasets while not exposing customer information creates various opportunities for health, life and group insurers.
Anthem, a large health insurer, partnered with Google Cloud to generate massive amounts of artificially generated medical histories, patient medical records, healthcare claims and related medical data so Anthem could scale and improve its AI systems.
In employee benefits, carriers could extract more value from their AI-powered underwriting systems by using similar artificially generated data.
Group insurers can produce more accurate rates and quotes that reflect the complexity and variability of real-world industry operations and employee health risks by feeding algorithmic underwriting solutions with synthetic data about past healthcare claims, medical histories, employee turnover rates and supply chain disruptions.
Of course, group insurers can't rely solely on synthetic data for underwriting. Real-world data, such as historical sales statistics, will always be valuable for automated sales and underwriting systems.
Tomorrow Belongs to Those Who Embrace AI
Insurers that embrace AI can gain a competitive edge by providing more personalized services, reducing costs, streamlining processes and enhancing productivity. As AI becomes more prevalent in the group insurance industry, we can expect to see more innovation and enhancements in how insurers operate.