AI in insurance augments the core professions rather than replacing them. Underwriters shift from keying standard risks to judging complex accounts and exceptions. Adjusters spend less time on intake and documentation and more on complex loss determination and empathy-heavy conversations. Actuaries move from data wrangling to model oversight and validation. The transition fails when carriers treat it as a headcount play and skip the reskilling and change management. The winners redefine roles, retrain people into the higher-judgment work the models leave behind, and give frontline staff a real voice in how the tools are designed, because a model the underwriters distrust gets overridden into irrelevance.
The work changes shape before it changes size
Insurance runs on expert judgment, and the professions that carry it, underwriters, adjusters, and actuaries, are exactly where AI lands. The near-term effect is augmentation, not elimination. A model handles the standard, high-volume, low-ambiguity work, and the human is redeployed to the complex and contested cases where judgment, negotiation, and empathy still decide the outcome. Straight-through processing might take 60 to 80 percent of standard submissions off an underwriter's desk, but the remaining accounts are the hard ones that need more skill, not less.
The industry also faces a demographic squeeze. A large share of experienced underwriters and adjusters is nearing retirement, and carriers have struggled to backfill that tacit knowledge. AI can help capture and scale expertise, but only if the workforce strategy treats the tools as a way to amplify scarce expert judgment rather than as a substitute for the training pipeline the industry underinvested in.
How three core roles change under AI
Plan the transition role by role, because the reskilling path differs for each. The table shows the shift and the capability each role must build.
| Role | Work AI absorbs | New capability to build |
|---|---|---|
| Underwriter | Keying and clearing standard risks | Complex-account judgment, model override discipline |
| Claims adjuster | Intake, documentation, routing | Complex loss handling, empathetic negotiation |
| Actuary | Data preparation and reconciliation | Model validation and governance oversight |
| SIU investigator | Manual claim scanning for fraud signals | Working model-flagged referrals efficiently |
| Service rep | Routine, repetitive inquiries | Escalation handling and relationship work |
Make augmentation real, not rhetorical
- Redefine each affected role explicitly, naming what the model does and what the human owns, so nobody is left guessing whether their job still exists.
- Fund reskilling into the higher-judgment work the models leave behind, especially complex-account underwriting and complex-loss adjusting where demand grows.
- Give underwriters and adjusters a real role in model design and a clear, low-friction override path, because tools the experts distrust get worked around.
- Use AI to capture retiring experts' tacit knowledge into decision aids and training, addressing the demographic gap directly.
- Communicate the change management story honestly and early, including what happens to roles, so the rollout is not undermined by fear and rumor.
How workforce transitions fail
- Framing AI as a headcount cut, which triggers quiet resistance and override behavior that neutralizes the model.
- Automating the easy cases without retraining people for the harder residual work, leaving staff underprepared for a heavier mix.
- Excluding frontline experts from design, producing tools that do not fit the real workflow and get ignored.
- Losing retiring expertise without capturing it, so the model inherits gaps the experts would have caught.
Measure the human side of the transition
- Model override rate by role, watching for distrust when it stays stubbornly high.
- Share of staff reskilled into complex-judgment work versus roles displaced.
- Time to competency for new hires using AI-captured expert decision aids.
- Retention of experienced underwriters and adjusters through the transition.
Frequently asked questions
Will AI replace underwriters and adjusters?
Not in the near term. It absorbs standard, high-volume work and redeploys people to complex accounts and contested claims that still require judgment and negotiation. The role changes shape, and demand shifts toward higher-skill work rather than disappearing.
How do we get experts to trust the models?
Involve them in design, give them a clear override path, and be transparent about how the model reaches its output. Experts who help shape a tool and can override it when their judgment differs will use it; those handed a black box will work around it.
How does AI help with the retirement wave in insurance?
It can capture retiring experts' tacit knowledge into decision aids and training, then use those to bring new hires to competency faster. This directly addresses the pipeline gap the industry created by underinvesting in training.
Related reading
Go deeper on this sector and topic.