Impact of Climate Change on Insurance

An AI perspective Amarnath Suggu

Global warming has adversely affected the climate, leading to more intense and frequent severe weather conditions worldwide that result in catastrophic losses for insurers. The year 2022 saw its share of extreme weather events—from floods in Australia and Asia to drought and severe winter storms in the United States and Europe. Natural catastrophes cost insurers nearly $121 billion in losses in 2022 compared to $116 billion in 2021.1 The numbers indicate a 5% to 7% increase in losses annually.

As studies note, insurers are finding it understandably difficult to forecast climate risks accurately. Unfortunately, the current risk models that depend on historical data do not consider the increased frequency and high intensity of present-day weather events, improved socioeconomic conditions and ever-increasing population in high-risk areas. This limits the ability to predict a true picture of the risk exposures and foreseeable losses. An actuarial challenge is finding a means to overcome this and stay ahead of the game.

Artificial Intelligence (AI) Can Predict Climate Change Risks

Data is collected from various global sources using Internet of Things (IoT) sensors to forecast weather. Traditional methods employ physics-based simulations that require solving complex mathematical equations to predict weather events. This approach uses enormous computing power, usually a supercomputer, and takes a long time to crunch the numbers. On the other hand, as a Forbes article notes, AI models bank on historical data to predict the outcomes rather than relying on detailed physics calculations. In addition, this approach accounts for physical processes that should have been considered in the simulations of traditional methods.

From my understanding, AI modeling can predict how a weather pattern will evolve or behave based on its resemblance to patterns that occurred in the past. It can recognize and understand similar patterns that might cause catastrophic weather events like hurricanes, severe winter storms and tornadoes. An AI-based approach requires much less computation than traditional methods. As a result, it is quick and light enough to be run on laptops. Due to its speed, many iterations of a particular model are run quickly with slightly different data points to determine the range of possible outcomes, also known as ensemble forecasting.

AI in the Insurance Value Chain

AI can be used in various stages of the insurance life cycle. For example, it can identify risks from climate change while underwriting new policies and expedite claim processing post-weather events. It also can provide timely alerts to policyholders about oncoming natural calamities and recommend safety precautions.


In my experience, AI-based underwriting is extremely quick, usually taking just a minute or two to generate a risk profile. In real time, it obtains the latest data from numerous sources—aerial images from satellites, property information from third-party risk providers, meteorological data from IoT sensors and loss data from historical claims—to arrive at the risk exposure. However, the true power of AI lies in its ability to quickly scan the images for risk hot spots and assess potential damages from the weather.

For example, The University of Washington has developed an AI model that uses meteorological data and past event history to predict weather and catastrophic events for up to a year for any given location. Zesty uses AI to calculate the proximity of water bodies, detect property elevation, identify the permeability of surrounding surfaces from aerial images and apply precipitation trends to predict the extent of flood damage from storm surges or heavy rains. InsurTech Kettle uses AI to predict wildfire risks and property damage. AI can flag wildfire risks based on the temperature trends and the type of vegetation surrounding a property.

AI models can combine potential damage information with the current labor and raw material costs adjusted for inflation to predict the losses that may incur from various weather events. Thus, they can provide a clearer picture of the potential weather risks and foreseeable losses, which can be used to arrive at the pricing.


Large claim volumes are a common sight after catastrophic weather events. Insurers often are flooded with a tsunami of claims that need to be settled quickly to win customer loyalty and avoid reputation damage. For example, more than 600,000 claims were filed in the aftermath of Hurricane Ian. It’s impossible to settle these staggering volumes without the help of technology. AI may aid in automating claim processes and expediting the settlements.

Accuracy-unproven AI-powered chatbots can record the first notice of loss (FNOL) from mobile devices. During catastrophic events, voice assistants and chatbots are deployed to handle call volumes and claim inquiries. Red Cross voice bots used in hurricane alerts are an example.

AI uses aerial images before and after a catastrophic event to identify the areas and properties affected. The model can detect the extent of damage from the image and estimate losses. For example, Geospatial Insurance Consortium (GIC) uses AI to generate damage assessment reports from pre- and post-event imagery for accurate and expedited claim settlements. Similarly, mobile apps like Flyreel and Upptec use AI to detect damage and estimate losses from images taken inside properties.

In conjunction with IoT, AI may be used to instantly settle parametric insurance claims commonly used to protect farms and residential properties from weather risks. For example, Descartes uses AI and IoT to settle parametric flood claims.

AI also assists in fighting fraud. It can quickly identify fraudulent claims and flag them for investigation. This is especially helpful during catastrophic events when some people try to take advantage of the situation and submit fake claims.

Risk Prevention

If we have timely knowledge of a storm’s occurrence and movement, we can minimize weather damage. Unfortunately, traditional methods of prediction are time-consuming and not so accurate. But with the advent of AI, many frontline organizations and InsurTechs (I list a few of them here) have successfully harnessed AI’s power to predict the occurrence and nature of weather elements. We can leverage the power of AI to limit the losses from tropical storms, floods and wildfires. Unfortunately, climate change highly influences these events and accounts for 70% of insured losses annually.

  • The U.S. Department of Energy’s National Lab has successfully developed a storm prediction model using deep learning. It can determine a storm’s path, how strong it can become and how quickly it can intensify. The tool provides an early warning and helps utilize relief efforts effectively.
  • Google’s AI-powered flood forecasting system can identify impact areas on Google Maps and indicate the water depth. It has been used in South Asia, issuing 100 million notifications and saving 360 million residents. In addition, Louisiana State University developed the Coastal Emergency Risk Assessment (CERA) tool, which uses AI to predict flooding and storm surges. Many state and national agencies in the United States use CERA.
  • Sonoma County in California has deployed cameras and paired them with an image recognition algorithm to detect wildfires. Descartes Labs uses AI software to detect changes in the thermal infrared data to alert the fire teams before wildfires erupt. In addition, Wifire Labs is using AI to map the rate of wildfire spread and direction based on vegetation dryness, topography and weather using data from multiple sources.

AI Will Be the New Normal

AI, it appears, is a good option for assessing climate change-induced weather risks. It can help insurers better understand and predict the risk of weather events and the accompanying losses, thus improving risk modeling. AI models are computationally less intensive and generate forecasts faster than traditional models. Many technology startups, research organizations and government agencies have successfully adopted AI to forecast and monitor natural calamities. I believe it is just a matter of time before AI replaces the traditional models as the de facto means in the fight against climate change.

Amarnath Suggu is a senior consultant in the BFSI CTO unit at Tata Consultancy Services Ltd. and is based in Chennai, India

Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries or the respective authors’ employers.

Copyright © 2023 by the Society of Actuaries, Schaumburg, Illinois.