Autoclaim automation is “not as close to completion as the industry once thought,” and “manual reviews and human judgment remain key to the estimation process,” says Mitchell’s parent company, Enlyte, in its 2022 Annual Trends Report.
Full automation should be considered “a long-term goal” as AI-assisted claims handling aims to live up to the hype that has built up around it, said Olivier Baudoux, Mitchell’s senior vice president for global product strategy and artificial intelligence. in chapter 4 of the report, titled “Claims Automation: Separating Fact From Fiction.”
Jack Rozint, senior vice president of repair sales for Mitchell, told Repairer Driven News that Mitchell believes “very firmly” in using AI in the estimation process, which he called “very viable and promising.”
The message to those disappointed by the technology’s failure to meet unrealistic expectations, he said, is “hang in there — things are going well, and there’s a lot more good to come.”
While the pandemic has accelerated progress in direct processing [STP]”There is still work to be done to deliver on the promise of a no-touch claim,” the report says. “As an industry, we are just beginning to reconcile initial expectations with the current state of technology.”
The report looks at changing expectations for the technology through the Gartner Hype Cycle, a graphic presentation developed, used and branded by the research firm Gartner to represent the maturity, adoption and social adoption of specific technologies. On that curve, the report says, expectations are between stage 3, disillusionment, and stage 5, mainstream acceptance.
“The key to this stage is realizing what is possible and how technology can be used in situations that deliver the most value,” the report said. Two areas where that applies, it says, are estimating and assessing and automating total loss claims.
The report seems to support long-standing concerns of the bodyshop industry. Wired magazine looked into the issue of photo-based estimates in April 2021 and found one overarching complaint among stores: damage detectable by a personal human appraiser is overlooked in photo estimations.
“While STP is a fundamental goal and certainly the long-term vision, it is equally important to recognize that manual valuations and human judgment remain key to the estimation process,” the report concludes. “In serious accidents or claims involving the latest vehicles, the involvement of an appraiser still brings tangible business benefits,
such as preventing cost leaks. Therefore, optimizing the core estimation solution to improve user experience remains a priority. This could be surfacing the right photos at the right time, calling in an appraiser when there is uncertainty about the selected part, or even asking a ‘real’ person to confirm what the machine might be unsure of.”
The report notes that while 97% of auto insurers surveyed recognize the value of STP, only 10% are making significant use of the technology, and less than 25% of global organizations have developed an enterprise-wide AI strategy.
As AI’s shortcomings became apparent, “sentiment in claims automation went from hopeful to hopeless,” the report said, with many organizations scaling back their STP investments from operational objectives to research and development projects. “They lost faith in non-contact estimates and instead began focusing on more straightforward and realistic AI use cases, such as assessing a claim or simply revising an already written estimate (including for subrogation).”
LexisNexis Risk Solutions, which recently reported that 79% of auto insurance companies were open to the idea of contactless claims, now predicts that telematics data and AI will lead to 60% of claims being handled automatically by 2025, the report said.
“[W]With only half of claims expected to be fully automated by 2025, this is now considered a long-term goal,” it says.
Mitchell’s report acknowledges some ways AI-driven estimates fall short, but notes progress is being made in a number of areas.
For example: “If you can’t decipher a vehicle identification number (VIN) and uniquely identify the model’s options, how can you accurately predict which parts need repair or replacement? Are you installing a bumper with sensors? What about headlights? And if the part is made of a special material such as aluminum, how should the repair plan differ?”
While AI could recognize “key” vehicle components, it became clear that “it had to learn hundreds or even thousands more,” the report said.
AI also hasn’t had the ability of human appraisers to contextually review all available photos before making a decision, or analyze multiple photos to determine unrelated past damage. On the other hand, human appraisers are “not ready to embrace their new role” in reviewing and approving pre-written estimates generated by AI, the report said.
Experience has shown that “while automation was designed to make claims handling easier, it was clearly unable to replace human appraisers,” the report said.
Mitchell argues that progress has been made in a number of areas. It says its proprietary AI solution, Mitchell Intelligent Damage Analysis (MIDA), can now predict more than 300 internal and external parts, and make recommendations for operations such as repair and replace, remove and install, and refinish and mix. Those are “consistent with OEM repair procedures,” Rozint confirmed.
MIDA is also capable of decoding VINs, largely by combining vehicle information with AI. “In order to make non-contact estimates to produce an accurate assessment, data on vehicles, repairs and historical claims are critical,” it says.
Recycled and aftermarket parts have been incorporated into the system and AI is now able to correctly identify the primary point of impact, as well as distinguish any unrelated previous damage.
The report suggests that STP technology is measured by gains on certain specific metrics. These include the number of predicted guess lines that are correct based on what is considered the ‘ground truth’, the number of parts that have been successfully mapped; the number of estimate lines that were added incorrectly and need to be revised, and the number of estimate lines that are missing or incorrect and require manual intervention.
On those metrics, MIDA has shown increases of 16% in the percentage of correctly auto-populated estimation lines, 26% in correctly recognized parts, 12% in correct identification of damaged parts and 8% in repair and replacement activities.
CIC panel delves into ‘myths’ of telematics
Wired: AI is coming to auto repair, and body shop owners aren’t happy
Featured image by Kwangmoozaa/iStock.