The insurance claims adjuster used to be a detective. They drove to accident scenes, photographed damaged vehicles, interviewed policyholders with practiced skepticism, and made judgment calls that could mean the difference between a family's financial recovery and ruin. The job required equal parts technical knowledge, interpersonal skill, and gut instinct honed over years in the field.
That job still exists in name. But the work itself has undergone a transformation so complete that adjusters who retired a decade ago would barely recognize it.
From investigator to orchestrator
The modern claims adjuster increasingly functions as an overseer of automated systems rather than a primary investigator. When a policyholder submits a claim through a mobile app — uploading photographs, describing the incident, providing documentation — machine learning models now perform the initial assessment that once required human expertise.
These systems estimate repair costs by analyzing damage photographs against databases of millions of previous claims. They flag inconsistencies in submitted documentation. They cross-reference weather data, traffic patterns, and historical fraud indicators to assess the plausibility of reported incidents. The adjuster receives a case file that has already been substantially processed, with a recommended payout and a risk score.
The human role has shifted from investigation to exception handling. Adjusters now spend their time on cases that the algorithms cannot confidently resolve — the unusual circumstances, the edge cases, the situations requiring genuine human judgment or the delicate interpersonal work of delivering difficult news.
The skills that matter have changed
This transformation has inverted the traditional expertise hierarchy. Technical knowledge of vehicle repair costs or construction materials — once the foundation of adjuster competence — matters less when software can access and analyze that information instantaneously. What matters more is the ability to recognize when the automated assessment has gone wrong, to understand the limitations of the models, and to navigate the increasingly complex conversations with policyholders who have already received algorithmic decisions they may not understand or accept.
The adjusters who thrive in this environment tend to be those comfortable with ambiguity and skilled at explaining technical systems to frustrated customers. They function less like independent investigators and more like translators between automated processes and human expectations.
Some in the industry describe this as deskilling — the replacement of hard-won expertise with button-pushing. Others see it as a genuine evolution, freeing adjusters from routine calculations to focus on the genuinely difficult cases that require human discernment.
The broader pattern
Insurance claims adjustment is not unique. Similar transformations are underway across professions where work involves assessing information, identifying patterns, and making probabilistic judgments. Radiologists increasingly review AI-flagged anomalies rather than scanning entire images themselves. Lawyers use machine learning to surface relevant documents from discovery rather than reading everything manually. Accountants oversee automated anomaly detection rather than examining every transaction.
The common thread is a shift from doing the work to supervising systems that do the work — with all the attendant questions about what expertise means when the intellectual heavy lifting happens in software you cannot fully inspect or explain.
Our take
The AI-and-jobs conversation too often gets stuck on a binary: will this profession be eliminated or not? The more interesting question is what happens to professions that are neither eliminated nor left untouched, but fundamentally reconstituted. The claims adjuster still exists, still draws a salary, still has a job title. But the job itself has become something else entirely — and whether that something else is better or worse depends entirely on what you valued about the original.
