Cognizant’s VP of Insurance, David Sexton, has forced a hard question into the boardroom. When a hiring request lands, directors now ask not only if a role is essential but whether AI could do it instead.
This is no longer an experiment. Sexton points out that much of what insurance companies call knowledge work has become pattern work. Tasks like claims processing, data entry, or risk scoring are rules-based and repetitive. Machines now do them faster, cheaper and more accurately.
But they do miss things around empathy and common sense from time to time. People, he says, should move to judgment, empathy and creative tasks to oversee the quality of outcomes.
Research from Cognizant and Oxford Economics estimates generative AI could add $1 trillion to the US economy by 2032 and affect nine in ten jobs.
Firms that froze hiring in 2023 are now debating what comes next. The default used to be “backfill when growth returns.” Now it’s “automate unless there’s a clear reason not to.”
Sexton writes that a typical European insurer runs a 500-person operation across claims, underwriting and customer service. Salaries alone cost about £25 million a year. Add benefits, offices, technology and management overhead and the annual bill rises to roughly £40 million. The cost climbs each year with inflation, while productivity remains capped by how fast people can process information and how consistently they can perform.
An AI-led model looks very different, he says. Deployment can cost between £80,000 and a few million depending on scale, but payback usually arrives within twelve months. After that, ongoing costs drop to about one fifth of a human-run system. Output improves each quarter as the models learn. Claims that once took days close in minutes. Customer queries get resolved instantly. Policy documents go out error-free. Compliance flags fall close to zero.
So the harder challenge is what happens to people - and this needs some thinking through. If AI takes the rote work, what is left for the humans who did it? Redeployment into higher-value roles? Most firms do not have enough of those roles ready. The risk is a hollowed middle layer - too skilled to automate, but too expensive for manual tasks.
As Guns and Roses said: Where do they go now?
What CEOs should do
- Treat automation as capability expansion, not just cost control.
- Leaders who frame AI as a headcount reduction tool will save money but culturally lose talent and trust. Those who use it to elevate human work will build speed and resilience. Look on it as augmentation rather than efficiency drives.
- Various laws of efficiency state people use all the extra capacity created by efficiency tech rather than bank the savings. It's possible augmented human productivity could drive enormous growth. Keep an open mind.
- Keep in mind, while many large organisations, such as UK government, have invested heavily in AI, few have reported its productivity gains yet. While industry will tell you to move, keep a close ear to the ground for gossip on who is actually getting the benefits.