Искусственный интеллект ускоряет обработку страховых случаев, но окончательные решения остаются за людьми

ORLANDO, Fla. — Artificial intelligence is helping workers compensation claims managers work faster and identify potential claim issues earlier, but human oversight remains essential, experts say. AI tools are being used to summarize lengthy claim files, assist with reserve reviews, flag potential litigation and identify claims that may benefit from earlier intervention, they said Monday at the Workers’ Compensation Institute’s Workers’ Compensation Educational Conference. The technology, though, is largely supporting rather than replacing decisions made by claims professionals, Bill O’Brien, workers compensation claims manager for the Troy, Michigan-based staffing company Kelly Services, said he uses AI to help manage claims across more than 40 jurisdictions, including researching state-specific workers comp requirements and providing a starting point for reserve and settlement evaluations. In one Virginia claim, an AI-generated settlement evaluation came within $10,000 of an attorney’s assessment, he said. The company settled based on the attorney’s figure, but the AI analysis helped Mr. O’Brien quickly get “into the ballpark” in an unfamiliar jurisdiction. “What is AI ultimately going to help us do? Manage our time and manage our days more effectively,” he said. That time-saving benefit is also evident in claim reviews. By using AI, the often-lengthy review process for “high-value claims” can be streamlined, said Michael Fountain, Palm Beach Gardens, Florida-based vice president of safety and risk management at telecommunications contractor Dycom Industries. Previously, preparing for discussions typically escalated across the company could take hours of reading claim notes. AI-generated summaries now give him a concise overview of a claim’s financial and litigation status before meetings with senior executives, he said. AI helps Victoria’s Secret sift through information to identify issues that require attention, said Jill Alfano, Syracuse, New York-based claims manager, risk and insurance, for the retailer. She said she uses AI to condense seven- or eight-page legal reports and highlight potential red flags when considering surveillance in workers comp and general liability claims. “That’s one real time saver,” she said. The company is also using predictive modeling to identify claims that may benefit from early case management intervention and those with a higher likelihood of litigation. If the model recommends early intervention and the adjuster agrees, a case management referral can be made without additional steps, Ms. Alfano said. “At the end of the day, who’s really winning is the injured worker,” she said, because the process can eliminate barriers to getting appropriate interventions. Still, AI is not making the final call. Even when technology flags a claim as likely to be litigated, “it’s still the people” who are developing risk mitigation strategies and working with injured workers to avoid litigation, Ms. Alfano said. Mr. O’Brien also cautioned that claims professionals must validate AI-generated information and the data supporting it. “You still have to inspect it. You still have to make sure it makes sense,” he said. Despite that caveat, claims professionals need to learn how to incorporate the technology into their work, Mr. O’Brien said. “If you’re not using AI in some capacity, you’re going to be left behind.” AI is also helping improve safety programs, Mr. Fountain said. Dycom is using predictive analytics on driving data to identify behaviors, such as speeding, that have been associated with future accidents, he said, adding that the company’s preventable vehicle incident rate has fallen to 2.33 accidents per 1 million miles driven from nearly eight in 2019.