Claims adjusters have won a title nobody wanted: the biggest AI haters in the American workforce. According to new research from Glassdoor, 98 percent of employer reviews written by claims adjusters that mention AI do so in the “cons” column — the highest share of any occupation the platform’s economists examined, and far above the 53 percent negative rate across all jobs. WIRED put a human face on the number on 31 August 2026, and the complaints are specific rather than abstract.
“Pushing AI to the point that you are asking humans not to use their thoughts and brains is such a turn off,” reads one representative review quoted by WIRED. “Stop forcing AI onto everyone,” reads another. Behind those lines is a profession that has lost a fifth of its jobs in a year, watched entry-level postings halve, and been handed error-prone artificial intelligence tools whose mistakes land on the desk of whichever claims adjuster picks up the phone next.
This article sets out what Glassdoor measured and how, why claims adjusters in particular have turned on the technology, the labour-market collapse underneath the sentiment, what insurers such as Lemonade and State Farm are actually automating, and what the episode teaches any organisation planning an AI strategy of its own. Every figure traces to a link in the References section.
Table of contents
- What Glassdoor Found About Claims Adjusters and AI
- Why Claims Adjusters Hate AI: The Errors They Clean Up
- The Labour Market Behind the Claims Adjuster Backlash
- Where Insurers Are Actually Putting AI in Claims
- Claims Adjusters Are the Extreme Case of a Wider Turn Against AI
- What the Claims Adjuster Revolt Teaches Anyone Deploying AI
- The Limits of the Claims Adjuster Data
- Claims Adjuster and AI FAQs
- References and Further Reading
What Glassdoor Found About Claims Adjusters and AI
The underlying study is “How workers feel about AI in 2026”, published on 27 August 2026 by Glassdoor senior economist Chris Martin. It flags a review as an AI mention when the text contains the standalone words AI, LLM, GPT, artificial intelligence or OpenAI, then classifies each mention as positive or critical depending on whether it appears under “pros” or “cons”. The detailed text analysis covers reviews from June 2025 through May 2026, from full- and part-time US-based employees.
The 98 percent figure
Within that dataset, claims adjusters stand alone. Martin writes that they are “nearly universally negative about their companies’ use of AI, typically complaining about AI-obsessed leaders forcing error-prone AI on them and their clients.” He told WIRED that he did not expect claims adjusters to hate AI quite this much — when he saw the results he did a double take, then started digging. The other most AI-critical roles are writers, journalists, accountants, customer service representatives, designers and IT staff, where critical comments outnumber positive ones roughly four to one.
How claims adjusters compare with everyone else
At the other end of the scale, company leaders are the most AI-positive group, with 67 percent of their AI comments appearing under pros. Sales and product managers, recruiters, software architects and data analysts and engineers round out the positive list. Even there the ceiling is low: positive comments outnumber negative ones only two to one, and software engineers — whose work is more exposed to code-writing tools — are 57 percent negative.
| Group | Share of AI comments that are negative | What Glassdoor says drives it |
|---|---|---|
| Claims adjusters | 98% | Leaders forcing error-prone AI on staff and clients |
| Arts sector | 89% | Employers’ approach to AI use |
| Nonprofits | 85% | Employers’ approach to AI use |
| Insurance sector | 81% | Concentration of AI-critical jobs such as claims adjusters |
| Writers, accountants, customer service reps | Roughly 4 to 1 negative | Replacement fear and forced tools |
| Software engineers | 57% | Code-focused role, high exposure |
| All jobs | 53% | Down from 81% positive in 2019 |
| Company leaders | 33% | Most AI-positive group, 67% under pros |
The gap between claims adjusters and the rest of the workforce is the story in one picture: every bar below is the negative share Glassdoor reports for that group.
Insurance as a whole is third most critical
The occupation drags its industry with it. Insurance ranks as the third most AI-critical sector at 81 percent negative, behind only the arts and nonprofits, and Martin attributes the insurance and telecommunications figures to “the concentration of highly AI-critical jobs like claims adjusters, writers, and journalists.” Company size matters too: AI comments are 51 percent negative at firms with fewer than 200 employees, rising steadily to 67 percent at those with more than 10,000 — and most claims adjusters work for very large carriers.
Why Claims Adjusters Hate AI: The Errors They Clean Up
The reviews are blunt because the failures are concrete. WIRED’s reporting by Kate Taylor traces the sentiment to a handful of recurring experiences, each of which puts a claims adjuster between a malfunctioning tool and an angry policyholder.
Initial loss reporting that misroutes claims
Take Ahmad Jackson. About a year ago he was working in the claims department of a major insurer when it switched initial loss reporting — setting up a claim and gathering information when someone first discloses an incident — over to AI. The promise was that simple claims would be streamlined and complex ones handed to people. Instead, Jackson says, he and other claims adjusters faced a sudden influx of misclassified claims that had to be rerouted to the right department. AI is “getting things wrong,” he told WIRED, “and it’s implementing more work onto the adjusters.” He quit not long after and moved to a different carrier.
Hallucinated summaries with real consequences
Jackson also encountered hallucinations while reading AI-generated claims summaries; when he inadvertently relayed those errors to claimants or their attorneys, he bore the brunt of their fury. Sandy Avina, a claims adjuster turned insurance consultant, told WIRED that claims adjusters do not have “a lot of faith in the output.” A smudge in a document from an attorney can produce a hallucination that leads to an incorrect payout; a missing point in an AI summary of a medical report can do the same.
One Glassdoor review quoted by Claims Journal put a number on it: “You often have to spend 15–20 minutes correcting AI mistakes because supervisors don’t verify anything.”
The customer blames the human
The cruellest part of the loop is attribution. Customers often do not realise that AI is the root cause of the confusion or misinformation, so they assume the claims adjuster made the mistake. Every hallucination becomes a reputational cost carried by an individual, not by the system. That is precisely the failure mode we described in our piece on enterprise AI agents and messy documents: the model is only as reliable as the scruffiest input it is handed, and in claims the inputs are photographs, scanned letters and hundred-page medical files.
| Where insurers deploy AI | What it is meant to do | What claims adjusters report going wrong | Who absorbs the error |
|---|---|---|---|
| Initial loss reporting | Set up simple claims, route complex ones to people | Misclassified claims needing manual rerouting | The claims adjuster |
| Claims and medical-record summaries | Pull key details from hundreds of pages | Hallucinated or missing points that change payouts | The claims adjuster, then the claimant |
| Photo and video estimates | Price damage without a site visit | No human check on a total-loss home | The policyholder |
| Chatbot claim filing | Replace the phone call to file a claim | Customers blame the adjuster for bot errors | The claims adjuster |
| Instant automated payout | Process uploads and pay within seconds | Works for simple claims, not complex ones | Nobody, when scoped correctly |
| Administrative “nuisance calls” | Extend a rental car by a few days | Nothing — adjusters say this genuinely helps | Nobody |
AI fatigue and the mandate problem
“There’s an AI fatigue,” Geoffrey Conrad, a claims executive in Mobile, Alabama, told WIRED. “We’re pretty much exhausted as far as the amount of AI being shoved down our throats.” One reviewer cited by Claims Journal said their employer was “tracking and forcing individual AI usage seemingly company-wide.” That matches Glassdoor’s broader finding that 14 percent of all AI-critical comments are about being force-fed tools. A mandate to use software that adds twenty minutes of correction to a file is not adoption; it is a productivity tax that leaders have chosen not to see.
The Labour Market Behind the Claims Adjuster Backlash
Sentiment does not sour in a vacuum. Martin’s conclusion, after digging into the numbers, was that the profession is in the middle of a reckoning, and the employment data support him.
A fifth of the jobs gone in a year
In 2024 the US Bureau of Labor Statistics projected that the number of claims adjusters would fall by 18,900, or 5 percent, over the coming decade, citing technology as a major force. Reality has run far ahead of the forecast. Between May 2025 and May 2026, employment in the sector dropped 21 percent according to BLS data cited by WIRED, while employment across insurance carriers and related activities fell just 2.5 percent. Claims Journal counts 13,100 claims adjuster jobs lost in the past year, a large slice of the 74,800 jobs shed by the broader sector, against a base of 65,700 claims adjusters employed in September 2023.
Entry-level postings have halved
The damage is concentrated at the bottom of the ladder. Glassdoor’s data, reported by Insurance Business, shows junior claims adjuster postings down close to 50 percent since early 2024, against a 15 percent decline for entry-level roles across the whole labour market. Total claims adjuster postings are 55 percent below their post-pandemic peak, compared with a 36 percent fall in postings overall. Senior claims adjuster postings, by contrast, remain about 80 percent above their 2017 level — a pattern consistent with experienced staff using AI to absorb the tasks juniors used to learn on.
| Indicator | Figure | Period | Source |
|---|---|---|---|
| Claims adjuster employment | Down 21% | May 2025 to May 2026 | BLS via WIRED |
| Insurance carriers and related activities employment | Down 2.5% | Same period | BLS via Insurance Business |
| Claims adjuster jobs lost | 13,100 of 74,800 sector losses | Past year | Claims Journal |
| Entry-level claims adjuster postings | Down nearly 50% | Since early 2024 | Glassdoor via Insurance Business |
| Entry-level postings, all jobs | Down 15% | Since early 2024 | Glassdoor |
| All claims adjuster postings | Down 55% from post-pandemic peak | To mid-2026 | Glassdoor |
| Senior claims adjuster postings | About 80% above 2017 | To mid-2026 | Glassdoor |
| BLS ten-year projection | Down 18,900 (5%) | Made in 2024 | BLS via WIRED |
| Carriers expecting to add staff / cut staff | 49% / 11% | Next 12 months | Aon and Jacobson study via Insurance Business |
Set side by side, the claims adjuster declines dwarf the equivalent figures for the labour market as a whole; each bar shows the size of the fall.
Layoff anxiety shows up in the reviews
“There’s no jobs, there’s no good job prospects,” Martin told WIRED, and Glassdoor’s data show that when workers sense layoffs looming, their reviews turn increasingly anti-AI. Across all occupations, a full 25 percent of reviewers who list AI as a con also mention layoffs — six times the base rate — and job insecurity and burnout keywords appear at 3.7 times the base rate. The insurance sector’s employee confidence, on Glassdoor’s index, has sat at an all-time low since May on a three-month average. For a claims adjuster, the tool on the desk and the redundancy notice in the inbox look like the same project.
Where Insurers Are Actually Putting AI in Claims
The technology behind the resentment is not hypothetical. Insurers have scaled AI from experiments to the front line of claims handling, and a few named examples show how far it goes.
Lemonade’s fully automated claim
Lemonade, founded in 2015, has promised from the start to replace bureaucracy with bots and machine-learning models. Its proprietary chatbot, AI Jim, handled initial reports 96 percent of the time by the end of last year, and automation handled roughly 55 percent of all claims, according to WIRED. A policyholder can upload photos, videos, receipts and documentation and, in the simplest cases, receive a payout within seconds. Lemonade spokesperson Paul Staats did not soften the implication: “AI is going to impact many jobs and threaten many incumbents in insurance and the economy,” he said, adding that Glassdoor’s report “should be taken seriously across the industry.”
Both Lemonade figures are shares of a whole, so the bars below show them directly.
Incumbents and the “mix of human and digital” line
More traditional carriers frame it differently. State Farm representative Justin Tomczak told WIRED that the company aims to “give State Farm agents and employees better tools so they can spend more time doing what matters most: helping customers,” and stresses that claims need a mix of human and digital expertise. Startups such as Liberate and Pace have raised millions on promises to “reinvent” insurance.
In practice that can mean a chatbot instead of a person to file a claim, photo or video analysis instead of a site inspection, and natural language processing that condenses hundreds of pages of medical records into a summary — the exact summaries claims adjusters say they cannot trust.
What a good use actually looks like
Not every claims adjuster rejects every tool. Jackson says AI is genuinely helpful for administrative work — what he calls “nuisance calls,” such as extending someone’s rental car booking by a few days. That is the shape of a use case that works: low stakes, easily verified, and taking a chore off a skilled person rather than a judgement. It is the same distinction we drew in our analysis of what AI leaves for human doctors: automate the paperwork around the expert, not the expertise.
Claims Adjusters Are the Extreme Case of a Wider Turn Against AI
The 98 percent figure is the headline, but the Glassdoor study’s broader finding is that the whole workforce is cooling on AI, and claims adjusters simply got there first.
From 81 percent positive to 53 percent negative
AI mentions in Glassdoor reviews rose 240 percent between May 2025 and May 2026, having already climbed 202 percent in 2023, 74 percent in 2024 and 164 percent in 2025. In 2019, 81 percent of reviews that mentioned AI did so only under pros. By 2026 that had fallen to 43 percent positive, 4 percent mixed and 53 percent negative. Positive mentions are still growing in absolute terms; critical ones are simply growing much faster.
What AI-critical workers actually complain about
Only one in five critical reviews is about being replaced. The detailed complaints are far more varied, and several of them describe the claims adjuster experience almost exactly.
Add the “pushers”, “distracted” and “delusional” buckets together and 35 percent of critical comments are about how AI is being deployed rather than whether it exists — the claims adjuster complaint in aggregate.
Women and Gen Z are the most sceptical
Sentiment also splits by demographics. Men’s AI comments are 45 percent positive against 32 percent for women; Gen X is 47 percent positive, Millennials 40 percent and Gen Z 33 percent. The gap widens with youth: Gen Z men are 42 percent positive but Gen Z women only 21 percent, and Glassdoor’s statistical controls find that job and industry explain none of that Gen Z gap.
| Group | Share of AI comments that are positive | How much job and industry explain the gender gap |
|---|---|---|
| Gen X men | 50% | Entirely |
| Gen X women | 42% negative | Entirely |
| Millennial men | 44% | About one third |
| Millennial women | 33% | About one third |
| Gen Z men | 42% | None |
| Gen Z women | 21% | None |
| All men / all women | 45% / 32% | Varies by generation |
AI opinions track opinions of the boss
The polarisation is stark. Only 23 percent of AI-critical reviewers approve of their CEO and 18 percent see a positive business outlook, against 86 percent and 81 percent for AI-positive reviewers. Senior-leadership ratings average 1.9 stars in AI-critical reviews and 4.0 in AI-positive ones. Workers who praise AI are 47 percent less likely to apply for another job in the month after their review; AI-critical reviewers apply at the same rate as everyone else, which suggests many claims adjusters are unhappy but not yet leaving — or, given the postings data, have nowhere to go.
What the Claims Adjuster Revolt Teaches Anyone Deploying AI
Few readers run a claims department, but the pattern generalises to any organisation putting AI in front of skilled staff. Four lessons fall straight out of the data.
Error cost lands on people, so measure it
Every hallucination a claims adjuster corrects is fifteen to twenty minutes of skilled time, plus the reputational hit when the customer blames them. A rollout that reports only “claims touched by AI” and never “minutes spent correcting AI” is measuring the wrong thing. Before mandating a tool, count the correction time, the reroutes and the complaints it generates, and compare them with the time it saves.
Mandates without trust produce 98 percent
“AI pushers” account for 14 percent of critical comments on their own. Forcing usage and tracking it, as one claims adjuster’s employer did, converts a tool into a surveillance programme. Proper change management means letting the people who will use the tool shape it, and being willing to switch it off when it fails.
Automate the chore, not the judgement
Jackson’s rental-car extensions and Lemonade’s simple, fully documented claims are the same category: bounded, verifiable, low-stakes work. Summarising a contested medical file or pricing a total-loss home from photographs is not. Well-designed business process automation starts from that boundary rather than from the vendor demo. If a company is exploring AI employees and autonomous agents, this is the scoping question to answer first.
Keep the keys with the human
Conrad’s line is the operating principle: “AI is just a tool. It should never be given the keys.” A claims adjuster who signs off an AI summary needs the authority and the time to overrule it — the same runtime-trust problem we examined in why authenticated AI agents still drift. Insurers demanding AI-era controls from their policyholders, as our guide to cyber insurance requirements in the UK describes, should hold their own claims tools to the same standard.
| What claims adjusters describe | Glassdoor complaint bucket | What a better deployment does instead |
|---|---|---|
| Tools mandated and usage tracked | AI pushers (14%) and AI alienation (10%) | Opt-in pilots with a kill switch and no usage quotas |
| Hallucinated summaries relayed to claimants | AI distracted (13%) | Source-linked summaries the claims adjuster can verify in seconds |
| Misrouted claims flooding the queue | AI delusional (8%) | Route only above a confidence threshold, measure reroutes weekly |
| Entry-level roles disappearing | Job killers (20%) | Redesign the junior job around supervision rather than deleting it |
| Customers blaming the human for the bot | AI distracted (13%) | Disclose AI involvement to the customer and own the error centrally |
| Nuisance calls handled automatically | AI users (41% of positive comments) | Start here, prove the saving, then widen |
The Limits of the Claims Adjuster Data
The numbers are striking, and they deserve the same scepticism claims adjusters apply to an AI summary.
Reviews are not a survey
Glassdoor reviews are self-selected; people write them when they are unusually happy or unusually angry, and AI mentions cluster in the most polarised reviews. Glassdoor itself notes that AI-critical reviews are also far more likely to mention layoffs and burnout, so some of the 98 percent is displaced anger at job losses rather than a verdict on the software. The report also acknowledges that it cannot always tell whether a generic “AI-obsessed leadership” comment reflects a bad implementation or the reviewer’s own prior view of AI.
Pros and cons is a blunt instrument
The positive-or-critical label comes purely from which box the comment sits in, and remarks under “advice to management” are excluded from that split. A claims adjuster who writes “AI is fine but the rollout was a disaster” under cons counts as fully negative. The sample is US-only, and the occupation-level figure rests on however many claims adjusters happened to review their employer in twelve months — Glassdoor does not publish the count.
The employment figures need care
Two of the labour-market numbers are reported slightly differently by different outlets: WIRED dates the 50 percent fall in entry-level postings from 2025, while Claims Journal and Insurance Business date it from early 2024. The BLS’s own projection is for a 5 to 6 percent decline over a decade, far gentler than the 21 percent one-year drop, and Aon and Jacobson’s survey found 49 percent of carriers still expecting to add staff. Long-term contraction is well supported; the exact pace is not.
Claims Adjuster and AI FAQs
Why do claims adjusters hate AI so much?
Because the tools fail in ways that create work and blame for them. Glassdoor’s 2026 report found 98 percent of claims adjuster comments about AI were negative, with reviewers describing leaders forcing error-prone AI on staff and clients. WIRED’s interviews add misrouted claims from automated loss reporting, hallucinated summaries relayed to claimants, and customers blaming the human for the machine’s mistake.
What exactly did Glassdoor measure?
Employer reviews from US-based employees that contain the words AI, LLM, GPT, artificial intelligence or OpenAI, classified as positive if the mention is under “pros” and critical if under “cons”. The detailed complaint analysis covers June 2025 to May 2026. Claims adjusters were the most negative occupation; insurance was the third most negative sector at 81 percent.
How many claims adjuster jobs have been lost?
BLS data cited by WIRED show sector employment down 21 percent between May 2025 and May 2026. Claims Journal counts 13,100 claims adjuster jobs lost in the past year, and Glassdoor’s postings data show entry-level claims adjuster roles down nearly 50 percent since early 2024 and all claims adjuster postings 55 percent below their post-pandemic peak.
Which insurers are automating claims?
Lemonade’s AI Jim chatbot handled 96 percent of initial reports by the end of 2025, with automation resolving about 55 percent of claims end to end. State Farm says it wants better tools for agents and employees while keeping a mix of human and digital expertise. Startups including Liberate and Pace have raised funding to automate parts of the process.
Is any AI use welcomed by claims adjusters?
Yes. Former claims adjuster Ahmad Jackson told WIRED that AI helps with administrative “nuisance calls” such as extending a rental-car booking. The common thread in accepted uses is low stakes and easy verification; the rejected uses involve judgement calls on contested documents or damaged homes.
What should an organisation take from this?
Measure the correction time AI creates, not just the tasks it touches; never mandate a tool staff cannot overrule; automate bounded chores before judgement; and keep accountability with a human who has the time to exercise it. For a broader view of where agentic tools are heading, browse the latest releases in our AI models and tools hub.
References and Further Reading
WIRED: You Know Who Really Hates AI? Insurance Claims Adjusters
Glassdoor Economic Research: How workers feel about AI in 2026
Claims Journal: Glassdoor — Adjusters Dislike, Fear AI More Than Others
Insurance Business: Entry-level adjuster hiring falls as insurers turn to AI
Business Insurance: Claims adjusters sour on AI
The Decoder: AI sentiment is turning sour as employee reviews reveal growing frustration
Allwork.Space: Workers are turning against AI at work, especially women
Axios: Worker confidence sinks despite positive labor market signs
Glassdoor Employee Confidence Index: Record low in July 2026
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.