Insurance Claims Adjusters Are the Loudest Voice Against AI in Their Own Industry

The people most critical of AI in American working life are not artists or writers. According to reporting by WIRED, research found that insurance claims adjusters who mention AI in their workplace reviews are critical of it 98 percent of the time, enough to make them the biggest AI sceptics in the workforce. What makes that worth reading closely is that adjusters are not speculating about a future risk. They are describing a tool they already use daily, which makes their objection a field report rather than a prediction. At Metaintro, we pay attention when the workers closest to a technology are the ones warning about it, because that is usually where the real implementation problems surface first.

What are adjusters actually objecting to?

Not automation as an idea. The complaint is narrower and more practical. Adjusters typically object to what the research described as AI obsessed leaders forcing error prone AI on them and their clients, and say that when the technology falls short they are the ones left to clean up a very human mess.

The mechanism matters. Ahmad Jackson was working in the claims department of a major insurer when his employer began using AI for initial loss reporting, the process of setting up a claim and gathering information when someone first reports an incident. The intention was to streamline simple claims and pass complex ones to people. What happened instead was an influx of misclassified claims that had to be rerouted, and AI generated hallucinations inside claims summaries. When he unknowingly passed those errors on to claimants or their attorneys, he took the anger personally. He left for a different carrier not long after, telling WIRED that the technology is getting things wrong and implementing more work onto the adjusters.

That is the shape of the objection across the profession. As a claims adjuster turned industry consultant put it, adjusters do not have a lot of faith in the output, because a smudge on a document from an attorney can produce a hallucination that leads to an incorrect payout, and a missing detail in a summarised medical report can do the same. Customers often do not realise AI caused the confusion, so they assume the adjuster made the mistake.

Why is that different from ordinary resistance to new tools?

Because the error lands on a person who did not make it. In most jobs, a tool that produces a bad output creates rework. In claims, it creates a wronged customer, an angry attorney, and a named individual holding the phone.

There is a second element that shows up clearly in the interviews. Geoffrey Conrad, a claims executive in Mobile, Alabama, described an AI fatigue, saying the profession is pretty much exhausted by the amount of AI being pushed at them. That exhaustion has a measurable cost in the hours workers lose supervising AI output. Fatigue is a different complaint from fear, and it is worth separating the two, because the remedy is different. Conrad's own position is not that the technology has no place, but that it should not be in charge, telling WIRED that AI is just a tool and it should never be given the keys. Jackson makes a similar distinction in practice, finding it genuinely useful for administrative work such as extending a rental car booking by a few days.

That is a more discriminating view than the headline sentiment figure suggests, and it matters for anyone managing an AI rollout. The objection is to placement and accountability, not to capability. The same distinction runs through the worker backlash against forced AI adoption and through the backlash that followed a surge in AI attributed job cuts.

What is happening to adjuster employment?

This is where the sentiment stops looking like grumbling. In 2024 the Bureau of Labor Statistics expected the number of claims adjusters in the United States to fall by 18,900, or 5 percent, over the coming decade. The actual decline has run far ahead of that projection. Between May 2025 and May 2026, employment in the sector dropped 21 percent according to BLS data.

For people entering the field it is sharper still. Entry level postings have fallen 50 percent since 2025. A profession losing a fifth of its employment in a year while halving its entry level intake is not adjusting gradually, and the BLS cited technology as a major force behind the decline.

That combination explains the sentiment better than any survey commentary. As the economist who analysed the review data told Wired, there are no jobs and no good job prospects, and when workers sense layoffs looming their reviews become increasingly anti AI. Some of what reads as hostility to a tool is really an accurate reading of a labour market. The wider job market backdrop is not reassuring either, with the Federal Reserve tracking how quickly AI adoption is spreading through US firms and putting work related use by workers at about 41 percent. The same contraction is visible from the employer side in the job cuts at Allianz attributed to AI adoption and in the financial sector shedding jobs at insurance carriers.

How far has automation actually gone in claims?

Further than most policyholders realise. In practice, automation can mean talking to a chatbot instead of a person to file a claim, having AI tools analyse photos or video to estimate a payout instead of sending someone to inspect a damaged home, or summarising hundreds of pages of medical records to pull out the crucial details. At the far end, an insurer's system can process uploaded photos, video, documentation and receipts and send a payout within seconds. AI startups have raised millions on promises to reinvent insurance while insurers scale their own use of the technology.

The official outlook has since caught up with what adjusters were already feeling. The Bureau of Labor Statistics now projects employment of claims adjusters, appraisers, examiners and investigators to decline 6 percent from 2025 to 2035, against a median annual wage of 78,000 dollars in May 2025 and about 21,600 openings a year over the decade. A declining occupation that still opens tens of thousands of vacancies each year is not one disappearing overnight. It is one where the entry point is closing while replacement demand keeps the existing workforce busy.

One company shows the ceiling. Lemonade has promised since its founding in 2015 to replace bureaucracy with bots and machine learning, and by the end of last year its chatbot handled initial reports 96 percent of the time, with automation handling roughly 55 percent of all claims. A spokesperson told WIRED that AI is going to impact many jobs and threaten many incumbents in insurance and the economy, and that automating allows employees to focus their empathy, care and expertise on the most complex claims.

Not every insurer is taking the same route. More traditional companies such as State Farm emphasise that claims require a mix of human and digital expertise, with a representative saying the aim is to give agents and employees better tools so they can spend more time helping customers. That divergence is the useful part for a job seeker. Two firms in the same industry are making materially different bets about how much of this work stays human, and which one you join determines what your role looks like in three years.

What does this mean for your career?

Five points, and the first applies well beyond insurance.

If you work in claims, treat the numbers as planning information rather than as background noise. A 21 percent employment decline in a year is not a cycle you wait out. The people best positioned are those moving toward complex, disputed and high value claims, which is precisely the work every party in this story agrees is staying human, from the insurer describing empathy on complex claims to the adjuster insisting the tool should not be given the keys.

If you are early in your career, take the entry level collapse seriously. Postings down 50 percent since 2025 means the traditional route in is closing faster than the profession overall is shrinking. That is a specific warning against betting on adjusting as a first job right now, and it matches the wider pattern of AI threatening the entry level job pipeline, which is what new graduates now have to plan around.

Two industry bodies are worth knowing while this is being argued out. The National Association of Insurance Commissioners is where state regulators set expectations for how insurers use models and automated decision systems, and the Insurance Information Institute publishes the sector research that gets quoted in most coverage of claims automation. If your employer tells you a tool is industry standard, those are the places to check the claim.

If you are considering a pivot, your adjusting experience is more transferable than it feels. Claims work is investigation, negotiation, documentation and difficult conversations under time pressure. Adjacent routes inside the sector start with the career insurance moves worth making and the insurance agent resume, and for a role built on difficult conversations, turning constructive criticism into career growth is more relevant than it sounds.

If you are anywhere else, take the accountability lesson. The core failure here is that a system makes an error and a named human absorbs the consequence. Before your employer puts a tool between you and a customer, the question worth asking in writing is who owns the output when it is wrong. That question is far easier to raise during a rollout than after a complaint, and the measurement follows the tool quickly, as it has where AI usage now feeds into performance reviews.

Finally, understand which half of the market you are in. PwC, analysing more than one billion job advertisements, found roles where human judgement is emphasised growing at twice the rate and with 42 percent faster salary growth than roles AI has made easier for non experts to perform. Claims adjusting is currently being pulled across that line in both directions at once, which is exactly why the profession feels unstable from the inside. That instability is compounded by the AI adoption gap and employer training failure, which is why there are specific questions worth asking before you take an AI heavy job covers how to test which side an employer sits on before you sign.

Why does an adjuster's own experience change how they see this?

Because many of them have been the customer, and that is the part of the story most easily lost in a productivity argument. Conrad understands that many people see claims adjusters as vultures circling after a catastrophe, but argues a good adjuster is both empathetic and determined to get the policyholder as much money as possible.

He knows because of what happened to him. Twenty five years ago, before he entered the insurance industry, his house burned down and was a total loss. His reaction to the idea of an automated assessment is worth quoting at length, since he said that if he had been told someone was just going to take photographs and generate an estimate from them, he probably would have lost his mind. What he needed at that moment, he said, was somebody to make sure he and his family were OK and safe. His conclusion is that AI can be helpful but not as a replacement for humanity, and that using it that way is a dangerous strategy.

That is a specific claim about what the job actually is, and it is testable rather than sentimental. A claim is filed on the worst day of somebody's year. The part of the work that looks inefficient from a process diagram, the call where nothing gets resolved and someone is simply reassured, is not overhead. It is a large part of what the customer is buying, and it is invisible in every metric an automation programme is likely to track.

This is also why the sentiment data is easy to misread as generational or reflexive. The people expressing it are describing a mismatch between what their job is measured on and what it is for. That gap is the same one that makes human judgment more valuable as AI spreads, and it explains why the human skills AI cannot replace are the ones being described here. Where a role's core value is the part that resists measurement, automating to the measurement tends to remove the value first. The adjusters saying the tool should never be given the keys are making that argument from inside the job, which is the most credible place to make it from.

What should employers take from this?

That the loudest critics here are the most experienced users, which is diagnostic rather than dismissible. A Lemonade spokesperson said the research should be taken seriously across the industry, and that is the right instinct.

Three things follow. First, error routing is a design decision, not an operational detail. If a system's mistakes reach the customer through a named employee, that employee is absorbing reputational damage for a decision they did not make. Second, fatigue is a rollout failure rather than a personality problem. The adjuster describing exhaustion at the volume of AI being pushed at them is reporting on pace and consultation, both of which are controllable. Third, the distinction workers are already drawing is the useful one. They accept the tool for administrative work such as routine calls and reject it for judgement. That is a workable operating boundary, and it came from the people doing the job rather than from a vendor, which is the case for treating AI productivity as something other than an automatic threat. It is also landing unevenly across the workforce, with a widening adoption gap between high earners and everyone else.

Related Articles

• A Trillion Dollar Wealth Firm Is Betting on Human Advisers Over AI
• Atlassian Cuts Customer Service Jobs and Denies AI Played a Role
• Commonwealth Bank Cuts More Jobs as AI Reshapes Australian Banking
• The Bank of Canada Says AI Will Lift Productivity Without Big Job Losses
• Agentic AI Is Set to Reshape Australian Finance Jobs
• AI Is Reshaping the Call Centre Landscape in the Philippines
• Which Customer Service Skills Transfer to Better Paid Roles
• Why Experienced Customer Service Reps Earn More, and When They Do Not
• Banks Are Posting AI Skill Jobs While Finance Payrolls Fall
• The Impact of AI on Banking Customer Service and Risk Management
• A Bank CEO Apologised After Calling Staff Lower Value Human Capital
• How AI Is Quietly Reshaping Half of US Work Hours

People Also Asked

Q: Why do claims adjusters dislike AI more than other professions?

A: Because the errors reach customers through them. Research found adjusters who mention AI in workplace reviews are critical 98 percent of the time, and the typical complaint is about error prone AI being forced on them and their clients, leaving them to clean up the mess. One adjuster described relaying AI hallucinations from claims summaries to claimants and attorneys, then absorbing their anger personally.

Q: Are claims adjuster jobs actually disappearing?

A: The decline is real and faster than projected. In 2024 the Bureau of Labor Statistics expected the number of US claims adjusters to fall by 18,900, or 5 percent, over the coming decade. Employment in the sector then fell 21 percent between May 2025 and May 2026, with the BLS citing technology as a major force. Entry level postings have fallen 50 percent since 2025.

Q: Is there any part of claims work that stays human?

A: Complex and high value claims, by broad agreement. An insurer heavily invested in automation described the goal as freeing employees to focus empathy, care and expertise on the most complex claims, while a more traditional insurer said claims require a mix of human and digital expertise. Adjusters themselves accept the tools for administrative tasks and reject them for judgement.

Future proof your career by tracking where the human work is actually moving in your industry. Metaintro follows AI adoption, hiring and pay across sectors so you can see which parts of your role are being automated before your employer tells you. Create your free Metaintro profile and find roles that still pay for judgement.