How Synthetic Data is Transforming Risk Assessment in the Insurance Industry

The accuracy and speed of determining the risk associated with the insurance of properties is being supercharged by the use of synthetic geospatial imagery.

The new breed of digital insurance companies, such as Lemonade and Hippo, is forcing the sector to turn to artificial intelligence and synthetic images to make faster and smarter decisions. 

Without the luxury of a national footprint, along with sales reps located throughout various countries, these new ‘virtual’ companies require the ability to conduct visual inspections remotely in order to assess property features along with their associated level of liability. For this, they are turning to geospatial imagery, such as satellite, aerial and drone images, which they analyze automatically using AI algorithms.

The main flaw with these intelligent algorithms, however, is the amount of images needed to train them. The more data they are fed, the more accurate they become. However, sourcing and annotating the imagery to teach these algorithms is difficult and incredibly time consuming. 

To train an algorithm to accurately detect objects it’s necessary to provide the algorithm with a huge quantity of images taken at different times of the day, in various weather conditions, geographic locations, etc. These images are then used to teach the algorithm what to look for and also need to be annotated in order to show the algorithm what each object looks like.

Historically these images were gathered and annotated manually, which is an incredibly long, slow and complicated process. This is because the algorithm needs to be shown how a pool, roof, or trampoline looks in every geographic location and in every weather condition, to enable it to accurately assess the condition of the property. 

The Power of Using Synthetic Data

This is why synthetic imagery is being used to supercharge the process, making it faster and more cost effective while providing more accurate results.

OneView’s synthetically generated images offer the ability to generate totally customizable images that are pre-annotated. These can be quickly used to produce accurate results. However, when these synthetic images are combined with real-world images, they have also been shown to drastically increase the accuracy of object detection. The algorithms can then be used to scan a predefined area to locate specific property features that could impact the rate of insurance. These features include the existence of a pool or trampoline, the condition of the roof, the material the roof is made out of, the angle of the roof, how close the treeline is to the home, and so on. They are also able to pinpoint issues that may not have been disclosed by the homeowner. 

In the insurance industry, where speed of the purchasing process and reduction in the level of involvement by the insurer and the customer are key, the unique advantages synthetic imagery offers can prove transformational. It not only brings better results from geospatial analytics, but also saves time in the collection and annotation of real world images to power machine algorithms. 

A Wide Range of Benefits

Next Generation Advances in AI are giving machine learning a radical upgrade, enabling synthetic data to supercharge image analysis and improve overall model accuracy. The result is that algorithms are becoming more accurate, answering more features, within a shorter time to market. 

As insurance companies embrace the power of synthetic imagery to power their AI tools, they can expect to increase the speed at which they assess risk, lower operational costs and ultimately provide better overall services to their customers. 

They can assess risk potential, receiving an instant insight into what kinds of potential claims might be filed for a given property, and speed up the ability to deliver insurance quotes. These quotes are also more accurate and more up to date as they can take into account changes to the property and updated property features. This also benefits their customers by taking the pressure off homeowners to try to accurately detail their property’s features, saving them time and hassle. 

Insurance Is Evolving, Don’t Get Left Behind

Digital insurance, remote visual inspection, and synthetic images, such as those offered by OneView, are all going to be significant drivers in the evolution of the insurance industry. 

Especially when you consider Accenture’s belief that, “leading insurers that use intelligent solutions to reinvent the customer experience and drive human-machine collaboration are achieving returns in excess of 10 times their investment in the technology.”

With new digital companies focused on disrupting the sector, legacy insurance companies are under pressure to adapt to a new competitive landscape. If they don’t react, they may learn the hard way and join the long list of companies, such as Nokia and Blockbuster, that didn’t heed the warnings and failed to innovate.