The insurance industry is undergoing a major transformation as artificial intelligence (AI) and machine learning (ML) revolutionise underwriting. Tasks that once consumed underwriters’ time, such as data collection and manual analysis, are now automated, enabling faster decisions and freeing professionals to focus on strategy and insight.

Leading insurers are positioning underwriters as ‘exponential’ professionals – not replaced but amplified by technology. Deloitte describes roles such as data pioneer, portfolio optimiser, dealmaker, and risk detective – underwriters who harness AI and alternative data to deliver deeper insights and real time foresight. These roles reflect a shift from hindsight (traditional risk scoring) to foresight (continuous, proactive portfolio monitoring).

Robotic process automation, advanced analytics, and cognitive technologies are now embedded in underwriting ‘workbench’ tools that automate routine inputs, simplify pricing models, and reduce time to quote from days to minutes. Meanwhile, augmented underwriting rules engines increasingly complement legacy systems – sometimes enabling ‘one question’ underwriting journeys that rely on permissioned alternative data rather than lengthy Q&A forms.

However, despite these tech advances, the human underwriter remains essential. As highlighted by professionals like Dina Tarantola Froner and Ludovic Proust, AI should be viewed not as a decision maker, but as a partner that augments expert judgement, especially in areas such as fraud detection or nuanced risk interpretation, where intuition and experience still matter most.

The impact of AI and machine learning on underwriting

As AI and ML reshape the insurance landscape, underwriters are finding themselves at the forefront of a data revolution. No longer consumed by repetitive manual processes, they’re now able to adopt smart technologies to enhance accuracy, agility and decision-making.

With this in mind, Dina Tarantola-Froner, Chief Underwriting Officer at International Medical Group (IMG), believes AI is best seen as a tool that complements – not replaces – human expertise: “Technology, such as AI, is a helpful, informative tool for underwriters when it comes to risk assessment and decision-making, but it shouldn’t necessarily serve as the ultimate decision-maker,” she said. “Rather, underwriters should look at AI as more of a partner that allows them to easily analyse data and information.”

AI’s ability to streamline complex data tasks allows underwriters to shift focus from operational to strategic thinking.

“AI helps underwriters efficiently look at data in ways that are less manual and burdensome, then respond to that data and apply it to the risk at hand,” Tarantola-Froner explained.

Ludovic Proust, Chief Underwriting Officer at Foyer Global Health, reflected on the contrast with earlier underwriting practices: “Underwriters, working on pricing for international private medical insurance (IPMI) products, used to spend unnecessary time gathering and cleaning data before even starting the pricing journey, with lots of analytical axis to be considered.”

Martin Leeks, Head of International Medical Malpractice at Carbon Underwriting, added: “Traditionally, underwriting has relied on historical data and manual risk selection. Now, intelligent systems can ingest vast volumes of structured and unstructured data … and predict a change to the level of loss severity across a portfolio.”

Ryan Husbands, Head of Underwriting at ERGO Travel Insurance Services, believes this to be the biggest impact AI and machine learning will have on underwriting.

However, in terms of automation, Tarantola-Froner noted that while AI offers clear advantages in areas such as fraud detection, it can’t yet replicate the intuition and experience of a seasoned underwriter: “AI is useful for automating some underwriting processes and is especially helpful for identifying fraudulent activity,” she said. “However, while AI can give underwriters the information asked of it, it’s not able to decide on that information in the same way an underwriter with years of industry knowledge and expertise can.”

Key data types helping underwriters predict claims and costs

AI’s promise hinges on high-quality, relevant data – an area where both traditional and contextual variables come into play.

“AI can help identify trends among a wide range of claims data, such as claim cost by benefit, common claim types, and policy deductibles – plus claimant data as it relates to age, country, state, etc.,” said Tarantola-Froner. “It can help underwriters more efficiently break down claims information to do a side-by-side analysis and easily identify trends.”

While AI can give underwriters the information asked of it, it’s not able to decide on that information in the same way an underwriter with years of industry knowledge and expertise can

This ability to dissect patterns is already feeding into IMG’s product development.

“Being able to conveniently analyse claims data and identify trends helps us determine claims drivers within our products and, in some cases, reshape a product offering to ensure our products are priced appropriately with proper deductible amounts,” she said.

For Proust, data sources need to be both wide-ranging and precise. “Historical medical data, demographic information, claims payment records, and turnaround times form the foundation for actuarial analysis,” he explained.

However, he cautioned that this foundation alone was insufficient. To accurately project costs, these data must be combined with complementary factors such as medical inflation, unemployment rates, age, occupation, education, access to medical care, and public health statistics.

Jack Farrall, Regional Manager for Verisk Life Health and Travel, agreed on the need for scope but stressed the importance of keeping data sets current and medically relevant. “Clinical data, claims history, and destination-specific health risks are proving essential,” he said. “Our in-house medical team updates Verisk’s risk assessment tools quarterly, ensuring that underwriting decisions reflect the latest medical insights and global health developments. This dynamic approach helps insurers anticipate costs more accurately and respond to emerging trends with agility.”

Husbands took a slightly different angle, noting that insurers didn’t always need to chase more data but could instead make smarter use of what they already have. “I expect the priority for most will be to use the existing risk and claim-level data collected in a more precise way to provide more insightful trend analysis,” he said. “There is a balance to be struck in maximising the amount of data collected while maintaining a seamless customer journey – the emphasis being to maximise the potential of existing data.”

Overcoming top challenges in tech adoption

Even as the potential of AI becomes clearer, insurers must overcome significant hurdles to embed this technology into underwriting processes, from technical infrastructure to cultural shifts.

“One of the biggest challenges faced by underwriters is simply understanding new technologies and how to best utilise them,” said Tarantola-Froner.

She also flagged the trust barrier: “Especially in the insurance industry, there is the challenge of customers being concerned about ‘a computer’ making decisions when it comes to their claim.”

Farrall noted the lack of consistency in how medical coverage is communicated and understood. “Disparate systems across distribution and underwriting often lead to under-declared conditions, which nearly 90% of UK travel insurers cite as a major pain point. The solution lies in industry-wide collaboration – standardising processes and improving transparency to better support customers.”

Further to this, Proust pointed to long-standing issues with legacy systems and data quality: “The biggest challenge is to access data from the legacy and outdated IT system. Data quality is also a major challenge as it might have been stored with limited information due to compliance and data privacy regulation. IT investment might be necessary to evolve in the data-driven environment.”

To meet these challenges, both companies and professionals will need to embrace structural change.

“Employers must provide change management support to assist underwriters throughout the transition,” Proust said. “In parallel, recruitment efforts will also focus on profiles with expertise in data science and analytics training.”

Tarantola-Froner emphasised the importance of education and transparency: “To overcome these challenges, it is first the insurer’s responsibility to gain a better understanding of new technologies like AI. With that understanding, both the insurer and customer can rest assured that AI has been shaped and programmed by industry experts with the knowledge to use it properly.”

Using tech for proactive and personalised insurance

With predictive analytics, insurers are no longer confined to reactive models. AI now makes it possible to proactively manage risk and personalise insurance at both group and individual levels.

“Technology helps provide underwriters with better reporting and analysis capabilities,” said Tarantola-Froner. “Not only does this allow underwriters to easily view risk at a macro level across an entire portfolio, but it also helps at the micro level, such as with individual accounts.”

She explained that this visibility enabled earlier intervention and more tailored coverage: “Underwriters can use this data to anticipate risks before they materialise... to segment insureds based on their lifestyle or risk profile and create more personalised insurance offerings.”

Farrall said to ITIJ: “By evaluating the stability and severity of medical conditions in the context of travel, we can offer fairer, more tailored coverage. AI and predictive analytics allow us to identify potential issues early and design insurance solutions that are not only reactive but proactive and customer-centric.”

Proust shared how this capability was already influencing the IPMI space: “Using AI and machine learning will enable underwriters, with the support of an actuarial team, to move to a proactive management of the risk.”

He added that this responsiveness was especially powerful in detecting and reacting to regional healthcare trends: “This technology should allow to detect earlier emerging trends in some countries and markets (e.g. spikes in chronic conditions, high-cost treatments in specific countries, unexpected inflation rates).”

Proust also highlighted the benefit of simulating financial impacts at the client level: “It enables tailoring benefits by client based on the current claims behaviour, by quantifying the cost impact of customised benefits and simulating portfolio outcomes to ensure financial soundness and pricing adequacy.”

“Communication both before and during travel to inform travellers about risks they may not be aware of is the obvious progression,” suggested Husbands. “It is important to show customers the value of the information being provided so it isn’t perceived as a way to increase marketing touch points.”

There is the challenge of customers being concerned about ‘a computer’ making decisions when it comes to their claim

Leeks highlighted how predictive analytics could identify policyholders at higher risk of future claims – not to penalise them, but to support better outcomes. For example, he explained, “if a healthcare provider shows unusually high rates of post-operative infections, targeted risk management support or additional training could help address the issue.” Similarly, “a surgeon facing recurring patient complications might benefit from focused training in a specific area before more serious problems arise and claims occur”.

Such proactive interventions not only help reduce loss costs but also create real value for insureds, shifting underwriting from a purely transactional process to a more relational, supportive one.

Essential skills for future underwriters in a tech-driven industry

As AI takes on more technical tasks, underwriters will need to adapt – not by abandoning their core skills, but by evolving alongside the tools they use.

Farrall told ITIJ that the underwriters of the future would need to be as skilled in data science as they are in traditional insurance principles. “Expertise in AI, analytics, and digital experience design will be essential,” he said. “Just as important, though, will be the ability to interpret complex medical data, work effectively across disciplines, and preserve the human touch in an increasingly tech-driven environment.”

“As the insurance industry becomes increasingly technologically driven, the top priority for underwriters should be understanding what emerging technologies can and can’t do,” said Tarantola-Froner. She agreed that “while adapting to new technology is important, underwriters need to maintain their skill set to remain experts of their trade... they can act as architects of new technology as it is being built and developed”.

She noted that staying informed was also key: “It is essential for underwriters to stay up to date on market changes and trends to apply this knowledge to AI technology. Ultimately, underwriters should be utilising AI as an aid that is able to simplify processes, enhance efficiency, and uncover new areas of focus.”

For Proust, critical thinking remains at the heart of good underwriting.

“I’ve always asked underwriters to challenge the tools output,” he said. “This will become even more critical with AI and machine learning. The risk is to miss the outliers and finally the risk appetite to become homogenised across the market.”

He added that future underwriters must blend deep insurance knowledge with technical fluency and teamwork: “They will have to challenge [AI’s] ability to interpret data and will also need collaboration skills to work closely with actuarial, data, medical, compliance.”

Conclusion

Underwriting is evolving fast, and it’s clear that AI and ML are playing a big role in that transformation. By taking care of the repetitive, manual work, these technologies are freeing up underwriters to focus on what really matters: making smart and strategic decisions to build better insurance solutions.

But while the tools are getting more powerful, the heart of underwriting hasn’t changed – it’s still about people, judgement, and experience. AI can spot patterns and surface insights, but it can’t replace the intuition of a seasoned professional or the ability to see beyond the numbers.

As Leeks pointed out, collaboration with technical experts is still key to aligning underwriting with broader risk strategies. “Even with growing automation, complex areas like medical malpractice still rely heavily on human judgement, context, and experience alongside data-driven insights.”