Transform commercial property underwriting with AI and NLU

Cyber is arguably the most extensive, expensive and difficult risk to underwrite that carriers have ever been faced with. Not only is the nature of cyber threat constantly evolving, making it extremely hard for organisations to prepare their systems adequately to resist attack, it is very difficult for carriers to make sure they are protected against unintended coverages that could prove very costly.

More and more, insurers are seeking to understand how they can design products that protect their clients’ businesses without leaving them over-exposed. Artificial intelligence (AI) and natural language understanding (NLU) are two tools that certainly have the potential to help insurers respond more quickly to threats as well as design policies more accurately, but they do need to be used in tandem with other strategies, including close industry collaboration and knowledge-sharing with clients.

To understand what avenues are open to insurers to improve their cyber resilience, Intelligent Insurer spoke to Pamela Negosanti, head of sector strategy, FSI at expert.ai. She reveals how industry-wide frameworks and more tightly-defined standards may go some way towards helping carriers and clients to have contract certainty.

This article is published ahead of the Intelligent Insurer webinar “Transform Commercial Property Underwriting with AI and NLU”, on April 29, 2021.

Why is cyber proving so difficult to underwrite precisely and what are carriers at risk from when it comes to policy ambiguity?
Policies are somewhat manuscripted—that is to say, they’re not universal all over the world. There is a great deal of variety in terms of wording and that’s the norm. But the problem with this is multifaceted.

On one hand, you can have wording that’s not crystal clear which makes coverage policies interpretable. That, in turn, can lead carriers to have unintended exposure. This is particularly the case with silent cyber and for emerging new risks. If cyber isn’t explicitly mentioned in the policy then it’s very possible the carrier could automatically be assumed to be covering it.

How should carriers be dealing with the ambiguities that leave them open to unintended coverages?
Carriers should be structuring their policies to add metadata. They also need to find ways to frame new perils better, and how they intend to group their policies accordingly. But, importantly, they need to communicate this to their customers and start thinking about shared knowledge.

What steps can carriers take to reduce their exposure to cyber risk, given there is so much that is yet to be understood about the threat?
A loss is a loss for everyone. No-one benefits. Rather than working in a system that is about detect and repair, we’re increasingly moving towards a predict and prevent model. Shared knowledge about how clients are prepared to meet risk is already common when underwriting for the property market, for example.

There’s another benefit for the industry in sharing knowledge across carriers. People don’t know specifically how to manage cyber because it’s new. It’s not something that executives have had years of managing. There’s no agreement on specific definitions and so on.

We need to have different carriers working together on defining some standards that can be agreed on. The idea of open insurance would be a benefit for everyone.

Finding some kind of global, universal standard is too big to tackle because of the different countries and jurisdictions involved. There’s a lot of variation. That said, there will be a need for some sort of framework so there can be a trade-off or compromise.

How much can automation and AI help mitigate the huge variety of cyber risks that underwriters have to consider?
Rather than automation, I’m thinking about augmentation in this process. Underwriting is an area where “no touch” is not feasible because it’s not providing the same level of expertise. I like the human part of combining technology: the human in the loop.

In reality, a lot of documents are not read at all in the building of a policy. You have just two or three days to process a new application. There’s simply not the capacity to ingest all that information. You need technology to support that and it doesn’t have to be 100 percent accurate because right now, with lots of documents going unread, you’re taking a blind risk underwriting that policy. The devil is in the detail and technology can get you there.

What do you hope attendees will take away from your webinar?
I would like people to take away the fact that there are great use cases and benefits applied to one of the major pain points in underwriting: the capability to understand the exposure of your portfolio. Cyber is the highest peril right now and it has been accelerated by the pandemic. We want to look at pioneers and what we can learn from them.

Join Expert.ai, Everest, Markel, Munich Re and Generali for the live webinar “Transform Commercial Property Underwriting with AI and NLU” (Thursday April 29, 2021 @ 3pm BST).  Register here now to join the discussion.