The insurance industry is well on its way to a future that takes full advantage of transformative artificial intelligence (AI) technology.
From using synthetic data in underwriting and pricing processes to incorporating new data into models to tackle climate risk, innovative AI technologies are delivering a host of new ways to work across the industry.
In a recent conversation with SAS strategic insurance advisor Franklin Manchester, we discussed four ways AI technology is altering how insurance works – and how it could reshape the industry’s future. Here’s a summary of what we uncovered.
1. Synthetic data remodels underwriting and pricing
Synthetic data, a form of generative AI (GenAI), is becoming a mainstream innovation engine in developing AI models for underwriting and pricing. This is not surprising considering that a recent study of global insurance decision makers showed 9 in 10 plan to invest in generative AI technology over the next year.
Historically, synthetic data was a parallel innovation engine. But more recent techniques – like transformer models, variable encoders, and advanced diffusion models – have significantly improved modeling efficiency and accuracy. Now, synthetic data is becoming a mainstream innovation engine for building models.
Synthetic data, for example, plays a crucial role in developing sophisticated models for new products like cyber or parametric insurance and rare events like natural disasters.
Insurers can take advantage of synthetic data to address several pressing data challenges:
- Bias and privacy. As a regulated industry, insurers must adhere to guidelines around fairness, transparency and consumer privacy. Synthetic data helps increase the representation of populations that are not fairly represented in models, while also anonymizing sensitive personal data and filling gaps in existing data.
- Cost. Acquiring massive amounts of real-world data can be expensive. Synthetic data is a cost-effective alternative that’s especially attractive for small and midsize insurance carriers.
- Innovation. Synthetic data levels the field by allowing smaller carriers to compete with larger ones. Insurers are using it for coding, price optimization, coverage recommendations and claims fraud detection.
Synthetic data plays a crucial role in developing sophisticated models for new products like cyber or parametric insurance and rare events like natural disasters.
2. New data helps combat climate risk
New data forms – from photos, email, streaming sensor data and mobile phone data – are estimated to create 400 trillion terabytes of data each day. Around 80% of this data is environmental in nature.
As climate risk evolves, insurers are using these new forms of data to proactively prevent or minimize losses and facilitate more resilient communities. Recent severe devastation and fires in California, for example, highlighted the need for insurers to respond effectively.
By incorporating a wide variety of new types of data, insurers can build more sophisticated climate risk models. In turn, they’ll be prepared to adopt a “predict and prevent” approach when they observe environmental changes.
Further, insurers need to act as climate risk advisors to businesses and communities, helping them respond to and recover from climate-related events. They can also deploy their capital in climate-benefitting projects and adaptation initiatives.
Insurers need to act as climate risk advisors to businesses and communities, helping them respond to and recover from climate-related events.
3. Fraud detection leans into new AI techniques and diverse data sets
Fraud detection has become more complex due to novel attack vectors like deepfakes and fake identities. It's estimated that a deepfake is committed once every five minutes. At the same time, the World Economic Forum predicts that by 2026, 90% of online data could be synthetic.
Insurers are prime targets of attacks.
One survey among UK insurers found a significant uptick in fraudulent claims since 2021. And with easy access to personal data and constantly evolving GenAI techniques, it’s easier than ever for fraudsters to submit fake claims.
Creating fake identities is the first step in setting up a ghost policyholder. The fraudster steals some real data from the internet and mixes it up with fake data to create a new digital footprint of an individual (the fake policyholder). After creating the identity, they can create fake photos and submit a false claim.
Machine learning and newer AI technologies that learn from patterns and respond rapidly provide the defense insurers need for this type of fraud. By incorporating diverse data sets with AI and GenAI tools, insurers can develop sophisticated, specialized models to successfully identify and thwart new types of fraud.
Here’s an example: Existing mathematical patterns (like Benford's Law) can be applied to images to identify inconsistencies in pixel patterns that indicate fraud.
4. Agentic AI shows high potential for underwriting and claims
Agentic AI systems are designed to act autonomously or in collaboration with users to accomplish certain defined goals – representing an advancement in GenAI applications. Rather than simply generating new content, these systems can autonomously plan and execute tasks toward specific goals.
Systems that plan, make decisions and take actions are gaining momentum in the insurance industry because they can streamline processes and improve collaboration and efficiency across functions.
For example, with claims automation, agentic AI can streamline the process by automating tasks and improving the accuracy of data extraction.
In the context of underwriting, agentic AI can:
- Streamline the placement process.
- Reduce delays.
- Improve collaboration.
For commercial underwriting, agentic AI shows tremendous promise during the intake process. This is where significant collaboration happens with numerous documents that have to be analyzed from a risk perspective.
Relying on a blend of natural language understanding and goal-directed behavior, agentic AI can help to extract data, triage submissions and dynamically populate models that assess risk for a policy or product. This can completely transform how insurers process and assess complex documentation.
By combining natural language understanding with goal-directed behavior, agentic AI systems can transform how insurers process and assess complex documentation.
Partnerships and an enterprise approach to AI
AI technologies including GenAI and agentic AI show high promise for significantly improving insurance operations and driving opportunities for growth – from underwriting and pricing to claims processing and customer service. As these technologies soar in value, other changes are afoot, too.
As they more fully embrace AI for multiple purposes across the enterprise, many insurers may also start to move toward public-private partnerships to address complex challenges like climate risk – more of an “ecosystem” approach.
I think we’ll start to see more public-private partnerships forming to solve complex problems. We need an ecosystem approach to handling these challenges.
Stay up to date with the latest insurance trends
