Canadian Underwriter

Insurers come out of COVID looking at AI-based claims tech

July 7, 2021   by Adam Malik

Print this page Share

Now that the global pandemic has expedited the use of artificial intelligence in auto insurance claims processing, carriers are left with a simple choice, according to one industry AI expert — invest or be left behind.

The promise of processing claims through AI has been some time in the making, said Olivier Baudoux, senior vice president of global product strategy and artificial intelligence for software company Mitchell’s auto physical damage division. Baudoux authored a report on the subject, entitled AI and Its Impact on Automotive Claims.

In his report, Baudoux observes that AI-enabled claims finally made their presence felt during the global COVID-19 pandemic, largely due to the government lockdowns and social distancing required to contain the spread of the virus.

Before the pandemic, fewer than 15% of auto claims were virtual, according to TD and LexisNexis; post-pandemic, the percentage of virtual claims shot up to between 85% and 100%. A more conservative estimate from LexisNexis Risk Solutions suggests 65% of auto insurance claims are AI-based.

Regardless of the size of the increase, it’s substantial enough to warrant insurance companies investing in new AI claims technologies, Baudoux wrote in a report for Mitchell. “Ultimately…organizations that embrace this ‘new’ technology to deliver a digitally driven claims experience…are best positioned to gain market share and consumer loyalty,” he wrote.

He used quotes around “new” because, as he put it, innovators have been trying to unlock the full potential of AI since the 1950s. But the auto insurance industry is only now starting to take advantage of such technology.

Opportunities around AI are “endless,” Baudoux said, be it detecting a car crash with Internet-of-things technology or processing a payment instantly following completed repairs.

For insurers, opportunities abound to use AI to automate the appraisal process and create touchless estimates.

“This can improve efficiency, shorten cycle time, and meet policyholder expectations for a streamlined, digital claims experience,” Baudoux said.

Baudoux listed four trends that put such an experience into reach: Shifting methods of inspection, the prevalence of big data, human-machine collaboration, and the growth of cloud computing and open ecosystems.

On the topic of shifting methods of inspection, Baudoux noted that virtual estimating used to be done only in low-severity claims. But the pandemic shifted inspection methods into the virtual world, leading to gains in estimate efficiency and consistency. “From images, appraisers can complete approximately 15 to 20 estimates per day [virtually], versus three to four out in the field,” Baudoux wrote.

Big data alone isn’t enough to reach the goal of making the claims process touchless, Baudoux argued.  “Access to a comprehensive library of vehicle, repair, and historical claims information is needed — along with the ability to quickly interpret that information using AI.”

Mitchell collects data on claim details and images, Baudoux said, citing an example. AI will then analyze the data and compare it to the company’s library of vehicle and repair info. Algorithms then create estimate lines for for appraiser review and approval.

Which ties into the need for human-machine collaboration. One can’t do the job without the other. So if the data is incorrect, the appraiser’s feedback can help teach the machines to make better decisions.

“As AI-powered solutions remove repeatable tasks, employees have more time to focus on complex claims that may require extra scrutiny,” Baudoux added.

Finally, cloud-based systems are required to access the data from anywhere. “Open ecosystems allow AI to easily access data, analytics and software across platforms and providers, giving carriers the ability to create a cohesive, end-to-end claims experience,” Baudoux wrote.


Feature image by

Print this page Share

Have your say:

Your email address will not be published. Required fields are marked *