Tech
Adopting AI in Insurance: What Changes, What Doesn't, and What to Do About It
Jane Damon | August 28, 2026
Footnotes
1 "Conning Releases 2025
Survey on AI and Insurance Technology: The C-suite Verdict," Conning,
June 25, 2025.
The insurance industry has been here before: There was a time when paper ruled everything. Policies could run hundreds of pages, at least two copies went to the broker so one could be forwarded to the client, and every policy was reviewed individually before a cover letter went out with it. Typing pools transcribed dictated letters, which were printed and mailed.
Fax machines shortened the time frame of delivery, then eventually email became the standard. When document management databases arrived, paper and electronic files ran side by side for years while technology caught up to the volume of business. None of it happened easily—systems were limited, upgrades did not always work, and everyone, whether they had been in the industry for decades or were brand new, had to learn something new.
Artificial intelligence (AI) is the next version of that same pattern: It is moving faster than digitization did, and it is reaching further into the work. It influences not just how documents move, but how decisions are made.
What AI Is Actually Doing in Insurance
AI adoption is no longer experimental. In Conning's third annual survey of US insurance C-suite executives, 90 percent of respondents reported being at some stage in generative AI evaluation, with 55 percent at early or full adoption. This is the strongest year-over-year increase of any AI technology survey tracks. 1
Ways That AI Is Used in the Insurance Industry
Underwriting teams use machine learning models to determine risk and price policies and language tools to pull risk information out of submissions, medical records, and broker notes. The underwriter can review the final analysis and make better decisions. In claims, tools are available to assess vehicle and property damage from photos, and fraud and documentation review are enhanced. A claims adjuster no longer needs to see the vehicle to make claims determinations. For safety, programs are available to capture photos of hazards, create documentation, and issue notices of noncompliance. Brokers are seeing the same shift in administrative functions, with platforms that automate reporting for wrap-up programs, certificate tracking, renewals, and policy form verification.
Most insurers are still behind in AI adoption, but it is coming, with them running pilot programs or partnering with established providers whose models are already built, shortening the time frame of implementation.
What AI Is Good At
AI is good at writing, editing, and explaining things—drafting reports, answering routine questions, formatting documents, and handling math and logic tasks. I recently took a Word document I had written and asked AI to turn it into a PowerPoint. It produced a 34-page deck, correctly structured and formatted to my preferences, and I only needed to adjust about five slides. Setting up the deck's structure, fonts, colors, and layout used to be tedious, but now it is almost instant.
But AI is only as good as what it is given. It makes assumptions and errors because it generates answers by predicting likely patterns rather than verifying facts. It can produce an answer that sounds confident but is simply wrong. It is not good at making judgment calls, which can create issues with underwriting and claims work. The output should be considered a draft, not a decision or a final product.
Why AI Cannot and Will Not Replace You
AI is a tool and tools augment work rather than replace the people doing it—at least not entirely. Job openings for finance and insurance have pulled back in recent years, even as the industry continues to have a shortage of experienced, knowledge-based talent. Roles built around data entry, transactional processing, and generalized tasks are the ones most exposed, since this work is what AI handles well; AI will not completely replace jobs, but it will enhance them, and workers will adapt.
For brokers and underwriters, things will shift, and AI will handle tasks such as routine quoting, policy reviews, and simple claims handling, leaving more time for complex risk management tasks, coverage disputes, and client relationships.
I use AI to read my work, including email and instant messages, and summarize my daily activity. It does not answer those emails; it prioritizes and highlights those it judges time-sensitive, and the final judgment is mine.
Are You at Risk? What Can You Do About It?
The honest answer is that it depends on your role. As noted above, administrative work, data entry, and transactional processing are genuinely exposed. The insurance industry will likely benefit from AI, with fewer hires as automation increases. The concern raised with the transition to AI is the career pipeline. Entry-level and administrative roles have historically been where people learned the underwriting and claims function, and AI cannot replace that kind of on-the-job training. If those roles disappear before new ones are built to replace them, the industry loses its training ground.
What you need to be doing is less about resisting the tool and more about getting ahead of it. When your company adopts a new AI-enabled platform, volunteer to be a super-user. Get trained with the first group, and you become the person others come to for help, not the person left behind.
Therein lies the opportunity—invest in the skills that are hardest to automate: reading an ambiguous claim, negotiating a coverage dispute, or managing a client relationship. Those are skills that will be harder to replace. Take any training your company offers; being a team player and learning will give you an early advantage.
Conclusion
We have adapted before: Typists became data-entry clerks, then document specialists. Fax operators became email correspondents. People managing paper files became document management administrators. In each shift, the job did not vanish; it moved toward the parts of the work where judgment matters. And the people who leaned into the tool early were the ones who ended up training everyone else on it. AI is the same kind of shift, just faster. The task in front of us isn't to compete with it, but to get good at working alongside it before someone else does.
Opinions expressed in Expert Commentary articles are those of the author and are not necessarily held by the author's employer or IRMI. Expert Commentary articles and other IRMI Online content do not purport to provide legal, accounting, or other professional advice or opinion. If such advice is needed, consult with your attorney, accountant, or other qualified adviser.