Buried away in the official documents published prior to the UK Government’s Budget fiscal statement last week, were some alarming estimates about the possible impact of AI on the  financial, insurance and professional services industries.

According to the Office for Budget Responsibility, which keeps an eye on economic policy and advises on the effect that government decisions might have, more than two million jobs could be replaced by AI across the sectors in the next decade.

Of course, that did nothing to sate the rabid drive on the part of the Chancellor of the Exchequer to be on the bleeding edge of AI modernity, but right on cue, on Budget day itself, German insurance giant Allianz warned of job cuts coming out of its adoption of the tech, with suggestions emerging that up to 1.800 roles “which are heavily reliant on manual processes today” might be replaced. That’s around eight percent of the current Allianz headcount…so far.

That said, the insurance sector is big into the transformation potential of AI, nowhere more evident perhaps than at Zurich Insurance Group where Group CIO Mario Greco is heading up the Zurich AI Lab, a new initiative aimed at advancing the use of AI within the insurance sector, backed up by newly-appointed Group Chief Transformation Officer Carlos Rey de Vicente and Group CIO and Digital Officer Ericson Chan.

Academic partners include University of St. Gallen professor Karolin Frankenberger and ETH Zurich professor Elgar Fleisch, who explained at launch:

ETH Zurich students are at the forefront of this new wave, turning ideas into impactful real-world AI applications. This lab creates the bridge between this talent and a leading business to build what’s next. Building on our deep expertise and cutting-edge research in business model innovation and incumbent disruption at the University of St. Gallen, we are excited to collaborate with Zurich, driving transformation, advancing new research, and contributing to society through impactful knowledge, innovation and partnerships.

Potential 

AI has produced a number of interesting results for business, according to Greco, but there’s a lot more that needs to be done:

AI is supposed to be more impactful, more revolutionary than this. So the question is how can we make this really visible, beneficial? And what is the size of the benefits that we can get from it?

Changes must be business-driven, he adds:

This is not a question of technology. Technology can do wonderful things, but if business doesn’t need them or doesn’t know how to use them, technology is useless. The purpose of the AI lab…is to develop and experiment business solutions that can be beneficial for business.

But AI is going to be a transformation shock to the system for the insurance sector, Greco admits:

Bear in mind that this is an industry having a business model which was set up centuries ago. Fundamentally, over the last centuries, this industry has replaced papers and font and pens with computers, but the business model has remained the same. Now possibly today, we can change and innovative business model, but this must be proven, must be proven right, and must be proven also with consistent application.

US experience

There are already exemplars of what AI can do for an organization such as Zurich, with the US mid-market providing some fruitful use cases, according to EVP of Middle Market US, Alexander Wells:

We touch over 10,000 middle market accounts, which is about 50,000 lines of business annually in the US. Our success is often predicated upon our ability to bring smart solutions to our clients and our brokers faster than our competition and with less questions. We know that our underwriters spend a lot of time collecting, summarizing information, waiting for data. And lastly, trying to reconcile that against our underwriting appetite and our underwriting guidelines.

Zurich has been using three AI-driven solutions specifically developed to address non-productive friction points in that process., he explains:

The work was done collaboratively with both internal and external AI specialists and with our underwriting community. It was developed, piloted and rolled out inside of the last 12 months. These are all live tools that we’re using now to make our underwriters and UAs more efficient. We expect that this is going to add 20% to our efficiency for both underwriters and UAs without having to change our customer-centric and consultative underwriting model. This is the future of what we will be doing in the United States by using AI.

Zurich is taking its own approach here, he adds:

There is a certain level of AI you have to start to embed into your business that I think everybody is doing. Again, I think we think about it a little bit differently in our middle market space than some of our competitors do in that I’m looking to enable smart decisions by my underwriters. I think some of my competitors are trying to make decisions for their underwriters using AI. I think that the secret sauce in the middle market is this combination – there’s individual account underwriting, there’s portfolio underwriting, and you need to give underwriters insights into both of those things. I need to give people a better understanding of the individual account that they’re working on and how that fits in the portfolio.’

Then the trick is whether those underwriters can combine that information through their own expertise to make good decisions, avoid bad accounts in good portfolio classes and write good accounts in maybe some struggling portfolio classes, right? That’s how you differentiate yourself from a sub-80s combined ratio versus 100 combined ratio, which is where the general market runs.

But overall, the direction of travel is clear not only for Zurich, but the wider insurance sector, he concludes:

I think everybody is still trying to figure out what the ongoing AI exposure and opportunity to provide insurances is. Right now, it’s all about how do we build it? And then we’ll see as that becomes a more mature operational exposure, how we participate in that?

It certainly starts at a large global corporate level that we’re going to have those conversations. But a lot of our business in the US. is tech business, and we see an increasing number of tech middle market accounts that are AI-oriented. They’re all shifting into this, and we’re understanding it as we go along, too.

My take

It’s worth noting – and chuckling over – that there is one aspect of the AI revolution that the insurance sector is less enamored of as leading firms like AIG, American Financial Group’s Great American, and WR Berkley are asking US regulators to be allowed to exclude AI liability from their policies, fearing the potential for damage and costly litigation that ill-considered AI adoption among their customers might lead to. If such an exemption is allowed, it could have interesting implications for AI tech providers who might find that responsibility for losses passes to organizations buying and deploying AI. While that potential risk of cost might incentivise enterprises to take care with governance, monitoring and testing before rushing in, it could also slow down adoption rates. A risk worth taking?