<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=318106&amp;fmt=gif">
Skip to content
9 min read

How AI is Reshaping Reinsurance Underwriting For The Better in 2024

Freeing underwriters from repetitive tasks and enhancing their decision-making capabilities, AI is generating strategic and competitive advantages for reinsurers.

From the mundane to the mind-blowing, the applications of artificial intelligence (AI) in reinsurance are expanding. For reinsurance underwriters, AI is beginning to deliver powerful benefits. Maximising, optimising, and analysing the information at underwriters’ fingertips to enable better risk selection and pricing decisions. Here is how:

1. Capturing, interpreting and ingesting data

A big problem in reinsurance is that the amount of information needed to accurately assess, select and price risks is getting bigger. A reinsurer’s success is increasingly being driven not just by the quality of its underwriters but also the data available for making underwriting decisions. However, no human underwriter can process and filter the volume of data needed to remain competitive using their brain alone.

Today’s reinsurance underwriters harness an ever-widening array of internal, third-party and public datasets to gain a holistic understanding of underlying portfolios and risks, much of which is in unstructured formats such as emails, PDFs and images.

AI is not only able to ‘read’ huge volumes of information that would take a team of human underwriter’s days or weeks, but also make sense of it in a way traditional IT cannot. Solutions already exist that can summarise documents or look for similar clauses in contracts. AI can understand not just the information on a PDF but also the context in which the words are used. For example, detecting positional relationships between words, tables, and other forms of data, can help identify what type of document it is and allow extraction of the most relevant information.

Reinsurers could potentially deploy this type of tool to read multiple broker submissions and extract relevant information from various sources. Extracted data can then be ingested into systems to create submissions, rate and assist in quote generation. Using extracted data, the reinsurer can also identify potential red flags around wordings, clauses, and risk appetites, and promote or demote risks to the underwriters depending on their relevance or attractiveness.

It is a bit like having an extremely dedicated underwriting assistant (with a truly terrible social life) to assist in speeding up the underwriting workflow journey identifying data to be input, used, and importantly analysed.



2. Analytical analysis

As well as capturing and interpreting data, AI can help reinsurers read between the lines and recognise trends in complex, multi-faceted datasets to offer game-changing insights.

AI is already used to interpret text to infer the emotion in an email, document or message. AI can detect the trends in pictures or video to highlight change (for example damage).

AI can be used to scour the thousands of contracts, security and safety records for every asset owned by a particular company, gather a holistic picture of its safety approach and factor this alongside other variables that might influence the attractiveness of the risk. The findings are then compressed and presented in a digestible format such as a simple risk score, helping to guide the underwriter’s decision on whether these assets represent a good risk.

Embracing this technology will allow reinsurers to take a deeper dive looking not only at the top few risks in an underlying insurer’s portfolio, as in the past, but beneath this by considering hundreds of risks and assets in the portfolio over a larger time period, with one eye on the characteristics they are keen to avoid. This scale and depth of visibility into underlying portfolios was incredibly time consuming before.

These tools are starting to be implemented today in complex commercial classes of business, and the data they analyse will increasingly be available to underwriters in real-time. We have come a long way from automatically adjusting motor premiums based on a handful of generic variables such as age and gender.



3. Intelligent decision-making

The next step in the AI revolution is to harness machine learning (ML) to make predictions and decisions based on a reinsurer’s experience and the bigger context of its business model and appetites, the (re)insurance market and the wider world.

An ML algorithm can, for example, compare a submission to other risks in the reinsurer’s portfolio and those that have been accepted and declined in the past to make an intelligent assumption on whether this is the kind of risk the reinsurer wants to write. The more data that is fed into the machine, the more accurate its predictions become.

Meanwhile, trends such as climate change, cyber risk and social inflation are pushing loss patterns into unchartered territory. Models based on historical data are increasingly challenged and we are certainly capturing data today whose relevance will only be realised when models develop further. AI can help us investigate and predict what the next drivers of reinsurance treaty profitability are going to be far quicker than a team of underwriters, with a day job, could ever do.

3. The future is already here

The main thing holding the industry back from accessing this deeper level of AI-powered analysis is experience, not technology. The algorithms and technology already exist but to learn and improve, ML relies on being fed vast amounts of information over time, and we are only beginning to feed these models now. AI also needs ongoing support, and people who blindly follow it will eventually meet a scenario in which the AI does not work.

This industry experience however is not just limited to training on huge volumes of data but understanding the results and ensuring the data being fed and the outputs are trustworthy. It is important to spot the good from the bad and feed this back ensuring a greater confidence in ML output in the future. This combined with tracking models lineage can aid in a greater understanding of how they evolve over time and help augment industry staff in what they do and the decisions they take.

For companies looking to gain a competitive advantage through AI and ML, it is important that they have an ecosystem (technology and people), and importantly a core platform, which can embrace and complement this technology and consume the information that is presented.

But to those biding their time in their adoption of AI – be warned. This is not a vision of reinsurance five or ten years from now; the world’s biggest reinsurers, with their vast data and analytics capabilities and teams of data scientists, are already building models designed to harness the full potential of these tools. And the data capture and trend analysis capabilities outlined above are already being implemented in complex reinsurance classes.

 The AI revolution is underway and the competitive gap between those who embrace it and those who do not will quickly widen. Therefore, reinsurers must consider engaging with this technology now to keep pace and maintain competitiveness. The robots are not coming – they are already here.


Subscribe To Reinsurance IQ

Stay updated with our newsletter containing the latest insights for reinsurance professionals.