What part will AI play in the future of commercial insurance pricing?

Commercial insurance pricing is evolving

‍The era of AI and advanced analytics is redefining traditional approaches to commercial insurance pricing. This transformation impacts the role of pricing actuaries, giving them the ability to answer a wider range of questions by investigating the portfolio experience, as well as improving the underwriting workflow. In addition, Large Language Models (LLMs) act as a virtual analyst enabling the automation of processes that were previously thought to require manual human effort.

Our pricing expert, Karol Gawlowski has shared some of his thoughts on the future of insurance pricing and how it is evolving so fast. Let’s find out what the future holds as the result of the ongoing analytical arms race.

Harnessing AI and Machine Learning models for smarter risk insights      

The future of pricing is not about finding a way to force LLMs in the process. As the amount of data collected by insurers grows, it becomes increasingly difficult to capture all the risk-determining patterns using standard statistical techniques, as fitting highly performant GLMs becomes a time-consuming task. Tree-based models – with gradient boosting in the lead –outperforms standard methods out of the box, and almost instantaneously delivers rapid and accurate predictions. Subsequently, experimentation with modeling beyond technical price becomes a viable option, providing opportunities to better understand and communicate the state of the underlying risk portfolio.

As the world’s first data scientists, actuaries have been dealing with numerical and structured data modelling for ages. In that context, the introduction of ML models can be thought of as an improvement. Now, thanks to LLMs, modelers can not only speed up the delivery and quality of pricing engine components but also, by accessing LLM APIs, extract signal from unstructured text data like policy wordings or claims descriptions; this automates operations such as text data cleaning that otherwise would require lengthy manual processes.

How does technology impact the commercial pricing landscape?

Insurers must act more like data companies to optimise the information available. As a result, for data-rich Lines of Business the interaction between actuarial and underwriting teams is likely to become tighter, more frequent and bidirectional. Now more of the information stored in historical records (either in a tabular or unstructured form) can play a role in enriching the underwriters’ expertise. In line with increased model complexity, actuaries (especially modelers) must adjust the way technical insights are communicated, so that the underwriters can benefit from becoming more technical, and appreciative of the quantitative insights.

Some specialty lines that are inherently too heterogeneous or lacking data (either in terms of the number of records or variables) may still see an improvement through automation of repetitive processes and portfolio monitoring.

The rising gap: how advanced analytics will define market leaders

The difference between players who manage to build robust pricing infrastructures will become more and more pronounced as time passes. Carriers capable of integrating advanced analytics into decision-making processes will dominate the market through efficient and accurate pricing reflective of the risk appetite, while others might struggle to keep up and remain competitive or profitable. It also means that better interaction with brokers and clients will be facilitated.

The hidden challenges of AI-driven pricing

Although it seems like the potential is only limited by modelers’ skills and creativity, there are some significant roadblocks along the way. Two concepts familiar to every data scientist are worth mentioning here:

·  The common modelling ratio where 80% of the time is spent on data preparation and only 20% on actual modelbuilding

·   The principle of "garbage in, garbage out."

In other words, organisations which romanticize about embarking on the AI and advanced analytics journey without first having their data capture, quality, and automated pipelines in check will struggle with training reliable models in general, let alone ML and AI.

Implementing novel techniques in some organizations might also result in culture shock, especially when it comes to communicating and trusting the results. ML models are not as easily interpretable as well-known statistical techniques, are more difficult to implement, and while large language models can outperform humans, the risk associated with their hallucinations deters organizations from including them in process automation, especially with external stakeholders.

What can actuaries do to prepare for the future?

Actuaries should allocate time to evaluate their current systems to ensure they have a robust and future-proof pricing infrastructure.

Optalitix isn’t the only pricing software out there, but it has been created and developed by a Corporate Actuary who understands the challenges first-hand. Therefore, the Optalitix platform includes advanced no-code solutions which are fast to implement, easy to adopt, and addresses user needs. We would welcome the opportunity to chat with actuaries who are interested in exploring their options.

Contact Us

‍