Forces reshaping underwriting, policy wordings, and reinsurance structures
Insurers are facing a rapidly evolving risk environment, shaped by the dual pressures of Artificial Intelligence (AI) and geopolitical instability. As these forces reshape underwriting models, policy wordings, and reinsurance structures, they are presenting challenges the industry has not encountered before.
FAnews spoke with Danny Joffe, Head of Legal at Hollard Group, to discuss how insurers are navigating these converging risks.
AI and geopolitical shocks: a growing exposure?
AI models are often built on historical data, but what happens when unforeseen geopolitical events, such as the Middle East conflict, fall outside those datasets?
Joffe said, “AI in my mind will be very good in dealing with regular standard risks, claims and investigations. And I am sure even the outliers such as war, political riots and catastrophic weather and health events will be built into the models who will in time become as good as dealing with it as human, for now human intelligence will deal with such outliers, but critical data related to the outliers will, in all likelihood, be better through AI than it would have been in the past.”
While AI models are effective for standard risks, Joffe emphasises that human intelligence is still needed for unpredictable events like wars or large-scale political riots. AI can improve decision-making by processing critical data, but the complexity of certain events will require human judgment.
Legal risks and complex policy wordings
As underwriting and claims decisions become increasingly automated, insurers are facing potential legal risks, especially when interpreting complex exclusions related to war and political violence.
“There is no doubt where the grey areas are reached, and there are many in policy interpretation being applied to the facts. AI claims investigation will have to improve before human intelligence can completely defer to AI models. We are currently in an interesting phase when it comes to AI, as we have seen significant advances over the last year, and I am informed that even more dramatic improvements are expected over the short-term. What is considered impossible for AI to do today will not remain so, and it will have access to hundreds of previous claims decisions (in many different insurers) and how they were applied, together with court cases and decisions by the NFO,” he said.
Joffe acknowledges the rapid advancements in AI but stresses that there will always be situations where human judgment is necessary. For now, while AI will access vast amounts of data to assist in policy interpretation, some claims will still require manual review, particularly those that involve complex or unprecedented circumstances.
Explainability of AI decisions in disputes
In situations where disputes arise, particularly in cases involving war or sanctions, how important will the explainability of AI-driven decisions become?
According to Joffe, “The dispute scenario will require a proper conversation between insurer and broker or insurer and client. War or sanctions decisions are not always that complicated to make, but where there are new issues that arise or where politically sensitive decisions need to be made, or where business calls need to be made, senior managers will still be relevant.”
Joffe points out that while decisions related to sanctions may be more straightforward for AI, complex or politically sensitive situations will still require input from senior management. Human decision-makers are necessary to handle the nuance of these situations, particularly when they involve high-level business or legal calls.
AI pricing and conservative reinsurance models
Could AI-driven pricing lead to a conflict with more conservative, geopolitically influenced reinsurance pricing? Joffe commented, “I think the AI models will not apply to all sections or product classes, but when they do, they will, in all likelihood, be able to adapt quickly. If they are not, then clearly where there are systemic dangers such as geo-political issues, serious weather pattern changes or things like COVID, human intelligence will always need to be able to make overriding judgment calls.”
Joffe suggests that AI will likely be able to adapt to new risks, but human oversight will still be required in cases involving systemic dangers like geopolitical conflicts or large-scale disasters. While AI can assist in rapid decision-making, senior underwriters will need to make the final call when situations deviate from the norm.
Real-time underwriting and legal constraints
Will AI enable more dynamic pricing or real-time underwriting for marine risks in volatile regions, or are there legal and contractual constraints that limit this flexibility? “AI will in time be more flexible than human underwriters in my view, if the models are robust enough. If wars break out or travel warnings are issued by certain governments. They will be linked into news and other sources to be able to immediately make decisions, whereas currently we find it needs to be referred to management, and often decisions are stalled until reinsurers get involved,” he said.
Joffe explains that AI could be more flexible than human underwriters in responding to geopolitical events like wars or travel warnings. By linking AI systems to real-time news sources, insurers could make immediate decisions, speeding up the underwriting process. However, legal and regulatory constraints may still prevent certain changes to policy wording until the proper notices are issued.
AI and sanctions compliance
With the evolving nature of sanctions regimes, could AI systems potentially lag in identifying prohibited exposures, thus creating compliance risks for insurers?
“As I said above, I think AI will act more quickly than the way it does now. If the systems are plugged into sanctions databases in real-time, the policies can be immediately picked up where there are breaches and clients cancelled where warranted. We are not seeing this today in the non-life space, specifically in the commercial space, where complicated corporate structures can be formed to evade suspicion,” he added.
Joffe sees AI systems as a potential solution to the current delays in identifying and managing sanctioned exposures. If these systems are properly integrated with real-time sanctions databases, insurers could act quickly to identify breaches, reducing the risk of compliance violations.
Aggregation risk in AI-driven portfolios
How should insurers think about aggregation risk, especially where AI-driven portfolios and geopolitical triggers combine to create correlated losses across multiple classes?
Joffe said, “Aggregation is probably not happening as it should currently, as one is not automatically putting together the various risks that may arise from the Middle East War, or a medical pandemic or the very severe weather storms that have become the norm in certain parts of the world. Travel insurance, marine insurance, liability cover and Guarantees are examples of events that occur overseas that can trigger real risks within South African Insurers.”
Joffe highlights the importance of better aggregation management, as risks arising from geopolitical events, medical pandemics, or severe weather can have far-reaching effects on multiple types of coverage. AI could help insurers track and identify these correlations, enabling faster and more effective responses to global risks.
Governance and oversight
Are insurers at risk of adopting AI capabilities too quickly, without adjusting their governance frameworks, policy wordings, and legal oversight?
Joffe said, yes. “I have no doubt that AI frameworks that respond quickly and in response to events happening immediately may well outpace the need to give formal legal notices, amend wording properly with the right oversight as demanded by PPR and soon COFI. There may be issues, but I have no doubt these will be built into the AI models that will alert compliance and legal to do what is necessary.”
Joffe acknowledges the risk of AI outpacing insurers’ governance frameworks. However, he reassures that these concerns can be mitigated by building the right checks and balances into AI models, ensuring legal and regulatory compliance without compromising the speed and efficiency that AI brings to underwriting and claims management.
Writer’s Thoughts
The intersection of AI and geopolitical instability is forcing insurers to rethink their risk management strategies. While AI has the potential to revolutionise underwriting and claims processes, human oversight remains essential, especially when dealing with complex, unpredictable events. By integrating AI with robust governance frameworks and staying ahead of regulatory developments, insurers can better navigate the evolving landscape of global risk. Please comment below, interact with us on X at @fanews_online or email me your thoughts.
