Here’s a number that should make you sit up: employment among insurance claims adjusters fell 21% in a single year. No mass layoff announcement. No industry crisis. Just AI in insurance quietly doing the work that used to require people. If you think your job is safe from this, read what happened to them first.
The profession that hates AI the most
Glassdoor dug through its own review data and found something striking: among claims adjusters who mention AI in their reviews, 98% are negative, according to Wired’s report. That’s the highest anti-AI score of any occupation in the American workforce. Journalists, teachers, artists, all less hostile than the people processing your home and auto claims.
Some of the reviews are brutally specific. “Pushing AI to the point that you are asking humans not to use their thoughts and brains is such a turn off.” “The AI apps this company uses are all trash.”
Why so angry? Because they’re the ones who see AI fail up close, and they’re the ones who clean up the mess.
The numbers behind the reckoning
The Bureau of Labor Statistics expected claims adjuster jobs to shrink by 18,900 positions, about 5%, over the coming decade. That projection already looks optimistic.
Between May 2025 and May 2026, employment in the sector dropped 21%. Entry-level postings fell 50% since 2025, according to Glassdoor. The standard career entry point, the way thousands of people started in insurance, is half gone.
This is what structural AI job displacement looks like in a real industry. Not a dramatic headline, just a year-over-year decline that compounds into a vanished career path.
For context on how widespread the pattern is getting: AI in insurance follows the same trajectory we’ve seen in other industries, where the numbers move faster than anyone projected. Robotics in warehouses, code generation in software, and now claims processing.
How insurers are actually using AI
You might think this is still theoretical. It’s not. AI in insurance is deployed right now, processing real claims today:
- Lemonade says its chatbot, AI Jim, handles initial reports 96% of the time, and automation processes roughly 55% of all claims end to end.
- Liberate, an AI startup building “reasoning agents for insurance,” raised $50 million.
- Pace, another agentic AI company for insurance, raised $46 million.
- State Farm publicly emphasizes mixing human and digital expertise, but even it is rolling out AI-assisted tools to agents and employees.
What does that look like in practice? Filing a claim by chatbot instead of calling a person. AI analyzing photos or videos of a damaged home and generating a payout estimate. AI summarizing hundreds of pages of medical records. In the most automated cases, upload your photos and documentation, and a payout gets processed within seconds.
That’s genuinely faster for customers. It also means the human role shrinks to whatever the AI can’t handle.
Why adjusters say it’s not working
The displaced workers have a specific complaint, and it’s not what you’d expect. They don’t say AI is too dumb. They say it’s confidently wrong, and the human gets blamed.
Ahmad Jackson worked claims for a major insurer until recently. His employer rolled out AI for initial loss reporting, and suddenly he was drowning in misclassified claims that had to be rerouted. AI hallucinations showed up inside claims summaries, and when he relayed those errors to claimants or their attorneys, he took the fury. He quit and switched carriers. His summary: “It’s implementing more work onto the adjusters.”
Geoffrey Conrad, a claims executive in Alabama, describes an “AI fatigue” across the profession. His line is worth remembering: “AI is just a tool. It should never be given the keys.”
Sandy Avina, an adjuster turned industry consultant, explains the failure mode clearly. A smudge on a document from an attorney can trigger a hallucination. A missing point in an AI summary of a medical report leads to an incorrect payout. And the customer doesn’t blame the AI. The customer blames the adjuster whose name is on the claim.
There’s a lesson here that applies far beyond insurance. When AI makes a mistake in a customer-facing process, the human in the middle absorbs the cost. That’s true in tech support, healthcare, banking, everywhere.
Where AI actually helps adjusters
To be fair, adjusters don’t hate everything about AI. Jackson says it’s genuinely useful for administrative work, like the “nuisance calls” that eat a day, for example extending a rental car booking. State Farm argues that better tools let employees spend more time on what matters: helping customers.
The pattern is consistent with every AI deployment that goes well. Boring, repetitive, high-volume tasks get automated, while the complex, emotional, judgment-heavy work stays human. Here’s the problem with AI in insurance: the automation arrived before the guardrails, and the humans got squeezed.
What workers in AI-threatened jobs can do
If you work in any role where AI is visibly taking over tasks, here’s what actually works, based on what’s happening in insurance:
- Learn the tools your industry is adopting. Adjusters who understand how claims AI works, where it hallucinates, and how to verify its output are more valuable than those who refuse to touch it. The worst position to be in is knowing less about the AI than your employer does.
- Move toward the oversight layer. Someone has to review AI output, catch the hallucinations, and own the decisions. That role exists in every industry rolling out agents. It’s usually better paid than the work being automated, too.
- Double down on what AI can’t fake. Empathy, negotiation, judgment under ambiguity. Conrad’s whole argument is that a good adjuster gets policyholders the maximum payout and makes sure they’re okay. Software doesn’t do that. The humans doing it well still have jobs.
- Watch the entry-level signal. If entry-level openings in your field are collapsing (like the 50% drop for adjusters), that’s the earliest warning sign. Senior roles linger while the pipeline dies, and then the whole ladder disappears.
If you’re thinking about a career change, our guide on what data center robots mean for jobs and our breakdown of why people push back on AI are both worth reading before you decide.
Is your job next?
Use the insurance story as a checklist. Ask yourself:
- Could a model do the most common task in my role with a 10% error rate, and would my company accept that rate?
- Is my work high-volume and repetitive, or does it involve judgment and relationships?
- Is my industry funding AI startups to do what I do?
One honest answer doesn’t doom you. Insurance shows that the middle of the curve gets hit first: entry-level roles doing standardized tasks. If that’s you, the playbook is the same as the adjusters’ best path, get ahead of the AI, learn it, and position yourself on the oversight side. And if you already use AI to manage your own money, the same caution applies: helpful tools still hallucinate. Our AI financial advice guide explains where that works and where it doesn’t.
Takeaway
The claims adjuster story is a preview, not a warning. AI in insurance is working, badly in places, but it’s working, and the jobs are going. The question for the rest of us isn’t whether our industry adopts the same tools. It’s whether we’ll be on the side reviewing the output or the side being replaced by it. Start learning the AI in your field this week, not next year.
