Insurers automating routine underwriting and claims work face a structural talent problem: the same automation that removes repetitive tasks also removes the training ground where junior employees build judgement. Nathalia Bellizia, a managing director and partner at BCG and its global leader for corporate finance and strategy in insurance, told Monte Carlo Today that companies must redesign jobs, training and accountability alongside their AI investments or risk burning out experienced staff while failing to develop the next generation of experts.
"We often think, rightly, that AI will remove repetitive, boring work and allow people to do the interesting, higher-value activities. What people sometimes do not realise is that, if you remove all the easy, repetitive work, none of it remains," Bellizia said.
Fewer cases, tougher decisions
Easy claims will increasingly move through straight-through processing, leaving people to handle unusual risks, stressful situations and difficult decisions. The work becomes more engaging but also more cognitively demanding. "There is a real risk of decision fatigue, burnout and, ultimately, attrition," Bellizia said.
A July BCG report illustrated the shift through a possible future underwriting model. An underwriter who today reviews hundreds of straightforward submissions might in the future work on 10 or 15 cases that are genuinely unusual, ambiguous or high-stakes. Concentrating human work on the most difficult cases creates a sustainability problem - employees no longer have routine work between those high-stakes decisions.
"As insurers embark on AI transformations, it is critical that they also design the human work," Bellizia said. "They need to create sustainable roles, supported by decision-support tools and clear escalation paths, and think carefully about workloads."
The expertise gap
Repetition has traditionally been insurance's training ground. A junior underwriting assistant might see hundreds of cases alongside an experienced underwriter, gradually learning to recognise patterns and recovering from mistakes. If insurers automate that work without replacing the experience it provides, they will need more expert judgement while developing fewer employees capable of exercising it.
"Strong human judgement will therefore become increasingly important, while that expertise will no longer be built through repetitive work," Bellizia said. "Expertise will be more valuable and, at the same time, scarcer and harder to build."
Insurers will have to become more deliberate about developing judgement. Bellizia pointed to structured mentoring, simulations, AI-enabled training and rotations through different parts of an organisation. She compared the approach with pilots learning to handle emergencies in simulators rather than waiting to encounter them in flight. "Unless expertise is built by design, it will not happen by accident, as it largely does today."
BCG estimates that change management, adoption and the redesign of talent and operating models account for roughly 70% of AI's potential value - tools and technology account for the remaining 30%. Companies that simply add AI to existing workflows may gain efficiency without fundamentally changing how decisions are owned or expertise is developed. This challenge sits at the centre of broader conversations about AI for Insurance and how AI Agents & Automation reshape professional work.
Accountability at scale
As AI takes over more execution, the people supervising it become accountable for decisions made at far greater scale. A single underwriter making a mistake on one file is one thing. An individual overseeing hundreds of straight-through processes handled by AI carries a different weight of responsibility. Insurers will need to define decision ownership across business, technology and risk.
Bellizia said small, senior, business-owned teams should monitor automated outcomes for quality, with the authority and expertise to intervene in real time when something looks wrong. The challenge is less about choosing between AI and people than about designing a system in which each adds the most value. "AI increasingly takes on scale, repetition, synthesis and routine decisions, while humans step in where judgement, ambiguity, empathy and accountability are important."
Why this matters for insurance professionals
For underwriters and claims professionals, the message is practical. The routine work that built your expertise is disappearing. The work that remains will be harder, and the path to senior judgement less obvious. Insurers that do not redesign how expertise is built will face a bottleneck: fewer people capable of handling the complex cases, and those few under greater strain. The companies that treat workforce redesign as a core part of their AI strategy - not an afterthought - will be the ones that keep their best people and develop the next ones.
Your membership also unlocks:
