Targeted, incremental adoption of AI can improve productivity

What does AI really mean in the context of reinsurance distribution?

We are at a very interesting point where a new paradigm of intelligence, both in understanding the market and helping with decision making, can sit alongside or on top of workflow applications.

But we are not about to suddenly see some all-encompassing AI that is going to do all the work for you.

This is still a people business, and that is not going to change radically any time soon, but being able to add tools that help practitioners in this area is very powerful. It is still early days, with huge potential, but there needs to be a balance of both optimism and paranoia.

What are the cost implications for the market?

As an industry, AI tools can certainly make people more productive and remove some mundane effort, which translates into real cost savings.

At an individual company level, insurers, reinsurers and brokers face challenges because there is so much current emphasis on investing in AI. Add to that how the scale of AI investment makes some other things more expensive, and this potentially disrupts other technology initiatives.

A simple example is the cost of memory chips. Data centres require huge amounts of memory to provide the computing power needed for AI training and inference. As demand for memory increases, so does the price of all servers and infrastructure upgrades. Some of this impacts investment in systems that are foundational for improved data management and automation.

Does AI replace or supplement existing technology?

Some tasks now done by humans will be replaced. Insurance and reinsurance involve many important but routine tasks that involve expensive people because real expertise and knowledge of market intricacies is required.

That knowledge can be captured with AI, creating opportunities across the value chain to make the workforce more effective and efficient, improve outcomes and reduce the cost of distribution.

Will it eventually replace people?

There is a shortage of people coming into the insurance market. That is particularly true in the intermediary space, where the demographics point to a shortage of knowledgeable people as a generation begins to retire.

However, those in the business will have to learn to use these tools. They will add significantly to productivity and allow some of the mundane effort to be replaced by technology.

Are there any challenges we should be concerned about?

Absolutely. At the top of the list are information security and data privacy, as more critical and confidential business data is used in AI models.

We can also expect a lot of discussion around regulation and governance, which is not necessarily a bad thing. We're still in the early days of this technology, and there is a lot of structure needed around it.

With a proliferation of experimental models, you can also end up with many competing models and approaches that each require onboarding, training, maintenance and security infrastructure.

These systems will eventually need to mature. It is the same pattern we have seen with other technologies, but this is happening in volume and at a significant pace.