Simon Glairy is a recognized expert in the fields of insurance and Insurtech, with a specialized focus on risk management and AI-driven risk assessment. Having navigated the transition from traditional actuarial modeling to machine-learning-driven underwriting, he possesses a rare perspective on how technological shifts reshape the labor market. Glairy is known for his advocacy of “human-in-the-loop” systems, arguing that while algorithms can process data, they cannot replace the seasoned judgment required for complex risk. His work frequently bridges the gap between historical economic patterns and the immediate structural challenges facing modern insurance carriers as they integrate generative AI into their core operations.
This discussion explores the long-term historical cycles of technological disruption, beginning with the 19th-century Luddite movement and moving through the automation waves of the 20th century. We analyze the current consensus—and significant disagreement—among leading economists regarding AI’s impact on productivity, job security, and wealth inequality. The conversation highlights the “insurance pincer,” a demographic and technological crisis where a massive retirement exodus coincides with the automation of entry-level training roles. Finally, we examine the fundamental shift in the insurance value proposition from a reactive “repair and replace” model to a proactive “predict and prevent” strategy, and what this means for the next generation of professionals.
History suggests that major technological shifts, like the Industrial Revolution, often lead to decades of wage stagnation even as productivity skyrockets. Why should we expect the AI transition to be any different, and what does a “four-decade” adjustment period mean for a professional entering the field today?
The historical record is quite sobering when you look past the glossy promises of technological boosters. Michael Strain of the American Enterprise Institute has highlighted a chilling reality: the Industrial Revolution eventually made everyone better off, but it left average real wages stagnating and the quality of life declining for roughly four decades. For an insurance professional starting their career in 2026, a forty-year transition isn’t just a statistical footnote; it is their entire working life. They could be hitting their mid-career peak before the labor market fully recalibrates to this new reality. We have seen this compression of time in subsequent waves—the automobile transition took about 25 years to settle, computer automation took 20, and the internet era about 15—but AI is moving with a velocity that suggests we might not have the luxury of a slow adjustment. When productivity jumps but wages don’t follow, it’s usually because the displacement of workers is visible and concentrated, while the new roles being created are diffuse and delayed. In 1812, you could see the weavers smashing looms in Yorkshire because their livelihoods were being destroyed in real-time, but nobody in that room could have imagined the millions of jobs that would eventually exist in global textile logistics or fashion retail.
The story of the Luddites is often dismissed as a simple fear of progress, but you’ve noted that they were right about what was being lost even if they couldn’t see what was being built. How does that “asymmetry of visibility” apply to the insurance industry’s current shift toward AI?
It is a visceral, sensory experience to lose a job. In the 1800s, it was the sound of the mechanized loom replacing the hand-weaver; in the 1980s, it was the silence of the typing pools as word processors took over. Today, in insurance, the displacement is happening in the quiet corners of claims adjustment and document summarization. We can point to 300 million full-time jobs worldwide that Goldman Sachs says are exposed to automation, and we can see the 700,000 manufacturing jobs that vanished in the UK during previous shifts. These losses are legible and headline-ready. What is harder to see are the “relationship managers” and “data storytellers” of 2035. When the automobile eliminated blacksmiths and carriage makers, it simultaneously birthed the motor insurance industry—a massive, entirely new line of business that didn’t exist in 1900. The challenge for us today is that the “reinstatement effect” is gradual. People like David Deming of Harvard remind us that people are the ultimate general-purpose technology, but that doesn’t help the claims processor whose “routine information-processing” role is being optimized away this morning.
A recent survey of sixteen leading economists, including Nobel Prize winners, showed a rare consensus on productivity but deep division on jobs and inequality. What does this lack of agreement among experts tell us about the predictability of the AI era?
When you have fifteen out of sixteen experts agreeing that AI will meaningfully boost labor productivity, you know the engine of the economy is about to get a massive upgrade. However, the fact that they split almost down the middle—five expecting job losses, eight expecting stability, and two expecting growth—should give us pause. It shows that even at the highest levels of MIT, Stanford, and Yale, there is genuine uncertainty. This isn’t just academic hedging; it reflects the reality that the displacement of a job is a technical event, while the creation of a new role is a social and economic one. Furthermore, with seven predicting wider inequality and only five thinking it will narrow, the “wild ride” Rebecca Henderson mentions becomes a political risk. If the gains of AI are “let rip” without institutional policies to share them, as David Autor warns, we face a future where the cognitive elite thrives while the rest of the workforce struggles to find a foothold. In insurance, this manifests as a divide between those who can govern the AI models and those who are simply managed by them.
We’ve seen previous waves of automation target physical or clerical labor, but AI seems to be aiming directly at the “cognitive” worker. How does it feel for the white-collar professional to experience the same pressures that blue-collar workers faced in the 1970s?
Justin Wolfers at the University of Michigan put it bluntly: white-collar workers are now having their “1970s moment.” For decades, the insurance industry felt insulated because our work was “analytical.” We thought reading a medical report, detecting fraud patterns, or drafting a legal summary required a human soul. But those are precisely the cognitive tasks AI excels at. The World Economic Forum’s 2025 report suggests 92 million jobs will be displaced by 2030, and insurance is in the high-exposure zone. It is a psychological shock to realize that your “codifiable knowledge”—the stuff you went to university for—is now a commodity. When 86% of employers expect AI to transform their business within five years, and 40% of core skills are projected to change in that same window, the pressure to reinvent oneself is immense. We are moving from a world where you are paid for what you know to a world where you are paid for how you apply judgment in the face of uncertainty.
You’ve described a “pincer movement” currently squeezing the insurance industry. Could you elaborate on how the retirement of 400,000 workers overlaps with the automation of entry-level roles to create a talent crisis?
This is perhaps the most dangerous paradox in the industry right now. On one side, the US Bureau of Labor Statistics tells us that roughly 400,000 seasoned insurance professionals are set to retire by 2026. On the other side, the entry-level roles—the “typing pools” of the past or the junior claims roles of today—are being automated away. These roles were the industry’s traditional training grounds. Joel Raedeke at Crawford & Company hit the nail on the head: AI is most valuable when it is used by an expert, but if you remove the junior roles where that expertise is built, you break the ladder. We have 91% of insurance and pension fund employers planning to hire for AI skills—far higher than the 62% global average—yet we are struggling to define how a new recruit becomes a “seasoned professional” when the algorithm is doing all the foundational work. We are essentially trying to build a roof while the foundation is being replaced by software.
If the industry’s value proposition shifts from “repair and replace” to “predict and prevent,” as some strategists suggest, how does the day-to-day work of an insurance professional change?
This is a fundamental redefinition of the profession. Historically, an insurance company was a giant “information processing” machine that waited for something to break and then wrote a check. Ajay Agrawal’s insight is that AI moves the needle toward prevention. If you can use AI to predict a leak in a commercial warehouse or a health crisis in a policyholder, your job is no longer about “adjusting” a claim; it’s about “intervening” before the loss occurs. This requires a completely different set of social and interpersonal skills. You aren’t just an actuary; you are a risk consultant. This shift is why 170 million new roles are expected to be created globally by 2030, even as 92 million vanish. The “new” insurance worker will spend less time on hand-drawn actuarial tables and more time on complex risk assessment for things that have no historical data—like AI liability or autonomous vehicle fleets.
In a world where AI can summarize documents and price standard risks in seconds, what specific human skills will command the highest premium in the coming decade?
The premium will shift toward what Daron Acemoglu calls “interpersonal and social skills.” As knowledge becomes codifiable and cheap, judgment and empathy become expensive and rare. We will see a massive demand for AI governance roles—people who can bridge the gap between actuarial science, data ethics, and regulatory compliance. We also need people who can manage “resistance” within organizations. Rafaella Sadun from Harvard Business School notes that the ability to build coalitions will be vital because AI creates internal winners and losers. If you are the underwriter who can price a novel climate risk where the algorithm has no “historical data” to learn from, you are indispensable. The same goes for the claims professional who can sit with a client at a moment of genuine distress. You can’t delegate accountability or human connection to a model, and that is where the future of insurance labor lies.
The insurance sector is essentially the “economic shock absorber” of society. What is your forecast for how the industry will handle its own internal disruption over the next five years?
The industry is currently in a race against its own legacy. My forecast is that we will see a “wild ride” marked by a sharp divergence between carriers. Some will “let it rip,” aggressively automating to cut costs, but they will likely face significant “ethical and reputational exposure” as they lose the human touch that defines the insurance promise. Others will treat this as a “workforce strategy” challenge, reinvesting the productivity gains from AI into retraining their staff for those 170 million new roles the World Economic Forum predicts. I expect that by 2030, the most successful firms won’t be the ones with the best algorithms, but the ones that successfully transitioned their people from routine processing to high-value strategic judgment. The algorithm is not the enemy, but as history has shown us from 1812 to the present, the industry that waits for total certainty before acting will find that the certainty arrives only after the market has moved on without them. The “prudent response” today is early, intentional action to ensure that the human general-purpose technology remains at the center of the risk equation.
