3 Ways Computer Vision is Transforming P&C Insurance in 2022
Have you heard the term “computer vision,” but aren’t quite sure what it means? Let’s look at what it is—and why it could prove critical to property and casualty (P&C) insurance in 2022 and beyond.
What is Computer Vision?
Computer vision is a field of artificial intelligence (AI) that enables computers to extract meaningful information from images and automate tasks that otherwise require human vision to process and analyze.
A wide array of factors have also accelerated the adoption of this technology within the insurance sector. This includes the dramatic and ongoing impact of the COVID-19 pandemic, a focus on speedy quotes and underwriting decision making, the scale required to handle natural catastrophe response, and increasing levels of fraud are chief among them.
Put together, these converging trends have created a business environment that is rapidly being shaped by disruptive new service models, evolving customer needs, and rising consumer expectations for speed, convenience, and personalized coverage.
To remain competitive, P&C insurers know they need to invest in technologies that help them understand, price, and mitigate risk effectively. Let’s look at three areas in which computer vision will play a critical role in the year ahead.
Underwriting: Instant Quotes, Remote Inspections Now Essential
Instant gratification is the name of the game for consumers and businesses looking for property insurance coverage through digital channels. The first insurer to deliver a compelling quote (usually) wins. But it can’t come at the expense of underwriting guidelines or inaccurate risk assessment. To meet this need, insurers are gravitating to a new generation of solutions that leverage computer vision to provide accurate, up-to-date property intelligence instantly and on-demand.
Today’s most robust solutions use this technology to analyze high-resolution aerial photography to extract property data on tens of millions of properties available instantly through APIs that can be integrated into consumer-facing online quote engines. Carriers like The Hartford and Kin rely on these services for property details like the size of a residential property, type of roof construction and its condition, the presence of solar panels, whether there are overhanging trees, a swimming pool, yard debris, or vegetation encroachment that could pose a fire danger—and more.
For homes and commercial properties requiring a ground-level inspection, computer vision is also employed in remote visual property inspection tools. This enables homeowners, commercial property owners, and others to do their own contactless property inspection via mobile app and the camera on their smartphone. Using artificial intelligence, solutions like FlyReel and Hover capture, catalog, and assess the interior and exterior of a property, as well belongings within the property—while saving on the cost of an in-person home inspection.
Not only do solutions like these help insurers understand the unique risk associated with insuring a property, but it also enables them to provide personalized coverage that’s priced accurately. It also helps them keep those customers long term. Today, 40% of customers will look for a new insurance company after filing a homeowners claim due to disconnects between the amount they thought they were covered for versus what the carrier actually covers. Now, technology can help eliminate these costly discrepancies.
Virtual Claims Reporting: From FNOL to OMG
Maybe it’s the Amazon Effect. After two years of distancing and a seismic shift in digital adoption rates, policyholders want claims handled fast. At the same time, the cost and risk associated with sending adjusters to conduct physical inspections continues to rise—especially in the wake of large-scale weather-related disasters like Hurricane Ida and this winter’s string of bomb cyclones.
Machine vision and deep learning-based neural networks are making a substantive difference here, too. Solutions like CAPE’s, for instance, can be critical in assessing the scope of damage due to catastrophic damage from weather-related events. By comparing pre- and post-disaster aerial imagery, it’s possible to accurately and efficiently assess damage at scale.
Meanwhile, mobile app-based solutions such as PLNAR and Matterport help accelerate first notice of loss (FNOL) for property claims by utilizing computer vision and augmented reality to automate low-severity claims operations and give adjusters the ability to virtually place themselves within the business or residence with accurate measurements and damage assessment.
According to JD Powers’ 2021 US Property Claims Satisfaction Survey, insurers offering virtual claims reporting over the past year have seen the highest overall satisfaction scores ever measured in the study’s 14-year-history. And a recent study from Capgemini finds that 80% policyholders who have initiated claims through image-based mobile apps have a positive impact on the customer experience, while nearly 30% say sending an adjuster actually hurts it.
Image Provenance: Avoiding the Wrath of Fraud
Speaking of claims: It’s no secret that fraud has increased in the age of COVID-19. With many families and businesses still facing tough times, a few will be tempted to stage car accidents, do damage to “non-essential” property such as sheds, barns, and separate garages or inflate legitimate claims losses. In more extreme cases, they just fake damages by using online or Photoshopped images. In some cases, otherwise law-abiding citizens will rationalize away these kinds of actions. They’ve paid premiums to those deep pocket insurance companies, why not get some cash back for a change?
In normal times, P&C insurers lose $40 billion a year to fraud. But in 2021, more than $80 billion has been paid out to bogus insurance claims in just the US. So it’s not surprising that 81% of insurers now use visual AI and other photo analysis technologies to authenticate claim damage, identify pre-existing damage, analyze digitally altered images, and index pictures submitted in other claims. That’s up from just 49% in 2018, according to the new State of Insurance Fraud Technology survey from SAS and the Coalition Against Insurance Fraud.
In 2022, also look for the increased use of computer vision in image provenance, which can be especially helpful in responding to natural disasters that may find policyholders and their families in need of rapid assistance. As it turns out, those same mobile apps used for underwriting applications can also ensure their authenticity. When policyholders use virtual reporting tools to document damage, computer vision and other technologies can give the insurer the highest confidence in image and video veracity.
This includes date, time, location, orientation (direction), and whether an image was synthesized using deepfake technology, according to PropertyCasualty360.com. It was critical to handling hundreds of thousands of claims during the deep freeze that swept across that led to $15 billion in insured losses throughout the Southwest in 2021. As the volume and severity of large scale claims events grows, so will the importance of this technology.
P&C: A Picture-Perfect Use Case
All of this is just the beginning. As technologies come to exceed human vision, delivering property appraisals with a 90% accuracy rate compared to 70% to 75% for manual inspections. And combined with additional data sources, machine learning, robotics, AI sensors, and other visual technologies will continue to create new opportunities for P&C insurance—an industry that is, after all, built on visual analysis and data.
In 2022, computer vision’s ability to help accelerate and enhance processes throughout the insurance lifecycle from underwriting, to claims, to fraud prevention already make it more than essential to an industry navigating a period of enormous change.
To learn how insurers can leverage computer vision-enabled property intelligence to select better risks, reduce expenses, and prove the customer experience, read the CSAA Insurance Group Case Study from CAPE Analytics.