We still need humans at the heart of what we do – even if some people do resist change. While some worry that artificial intelligence is sounding the death knell for many careers, others see it as a tool to make life easier for insurance professionals.
“I am definitely more optimistic,” said Pravina Ladva, group chief digital and technology officer at Swiss Re. Mass layoffs elsewhere are being conflated with artificial intelligence, she argued.
“People are putting two and two together and making a hundred,” she said. “We actually have more work, more analytics, and more responsibilities than we have people to manage them.”
Across the sector, the tension is less existential than operational. Carriers are wiring AI into claims triage, pricing workflows, and knowledge capture. Brokerages – especially in commercial lines – are inching toward more structured submissions, automated comparisons, and cleaner handoffs to markets. The tools exist; the bottleneck is human.
From the front lines of distribution, Shane Bennett (pictured), a director at the Canadian platform Quotey.io, puts it bluntly: “Brokers basically stick to what they know.” The friction is behavioral, not technical. “Changing habits kind of feels like more work than it's worth to them,” he said. In Bennett’s view, the resistance isn’t anti-innovation so much as a defense of the day’s cadence: “They're kind of more anti-disruption in their day.”
That day is crowded. Retail and wholesale brokers still juggle multi-market submissions, uneven portal experiences, and document ping-pong with clients. Generative AI is already being pointed at the “glue work” between these steps: normalizing intake data, checking applications for completeness, extracting terms from offline quotes and organizing side-by-side comparisons. “AI is basically going to handle the grunt work, the data entry, the comparisons of quotes, the back and forth that eats up broker's day,” Bennett said.
If this sounds like incremental plumbing, capital markets research suggests the underlying capability is anything but. An M.I.T. case study by Andrew W. Lo and Jillian Ross noted, “An LLM (Large Language Model) can role-play a financial advisor convincingly and often accurately for a client,” while warning that “even the largest language model currently appears to lack the sense of responsibility and ethics required by law from a human financial advisor.” And in a separate study out of the University of Chicago Booth School of Business, the authors reported GPT-4’s accuracy on earnings-direction calls was “remarkably higher than that achieved by the analysts.” The same paper went further on tradability: “We find that the long-short strategy based on GPT forecasts outperforms the market and generates significant alphas and Sharpe ratios. For example, alpha in the Fama-French three-factor model exceeds 12% per year.”
For insurance professionals, those findings are a double-edged marker. If general-purpose models can outperform in pattern-heavy finance tasks, carriers will push them deeper into underwriting support, exposure analysis and portfolio steering – while compliance and model-risk teams race to harden guardrails. On the distribution side, brokers will feel pressure to prove where the human remains essential: articulating risk, negotiating with markets, and translating choices for clients under real-world constraints.
Bennett sees the winners emerging on that front. “The ones that will win won't be those who send out submissions the quickest, but those who know how to translate those complex risks into simple, confident decisions for their clients,” he said. Even in a more automated front end, he expects small businesses to prize human guidance at the moment of commitment. In his words, “they want that personal touch but they also want quotes and they also want good prices and they also want to know what they're getting into.”
Ladva makes a complementary argument from the carrier vantage point. With a seasoned workforce edging toward retirement, she said AI is a mechanism for institutional memory – codifying underwriting judgment so the next cohort isn’t starting from zero. The priority, she added, is execution with the people who will use the systems every day. “Technology alone doesn’t create value,” Ladva said. “It’s the combination of data, tech, processes, and colleagues that will transform insurance. That’s why the human has to remain at the heart of everything we do.”
There are, of course, limits. Consumer tolerance for disembodied service is uneven. “What we learned, though, was most people who are consulting these resources are verifying what they hear with a financial advisor,” said Kevin Keller, chief executive of the CFP Board, in a separate discussion about advice tools. Insurance buyers – particularly in commercial lines – are likely to behave similarly: start with digital convenience, finish with a human’s assurance that coverage matches risk.
That hybrid future carries practical imperatives for leaders:
Tighten the data spine. AI’s lift in submissions, comparisons and endorsements depends on clean, structured intake and consistent document handling. Sloppy inputs still produce messy outcomes, now at scale.
Redesign work, not just tasks. If automation shortens the time to a quote, reallocate the saved hours toward risk discovery, client education and renewal strategy – areas where machines are assistive, not decisive.
Invest in change management. Adoption isn’t a memo. Pair frontline brokers and claims handlers with technologists; treat them as co-designers, not end users to be trained after launch.
Double down on judgment. As model-assisted pricing and triage spread, the premium on human negotiation, market sense and client trust will only rise.
For all the headlines, what’s taking shape in insurance is less a displacement story than a productivity contest. The firms that bend workflows around new capabilities, and the brokers who anchor them in client outcomes, are likely to move fastest. The technology is arriving either way. The question is whether the industry can change its habits quickly enough to keep up.
