In brief
- Ecclesiastical insurance is cautiously optimistic about AI. It does a lot of data crunching and routine processing, so there’s certainly an opportunity for AI to generate value. Beyond that, the most immediate benefits are in decision support, supporting underwriters as they process masses of data, looking for those crucial patterns that inform the best decisions
- Technology provider partnerships must change. Insurers must engage with specialists who can guide and support them on the digital transformation journey. Which is fine, but the days of replacing a massive core system with something that will, hopefully, transform every process are more or less behind us. Timelines are shrinking, so we need to be able to move faster, respond quicker and integrate with external partners while operating in both new and traditional worlds
- We talk glibly about the importance of collecting and using data in the future. The fact is, we're already producing masses of data through AI and machine learning. Deriving the benefits of that data in a quality way that supports the business and future activities is the hard part
- Claims handling is changing massively. A lot of the data collection grind is disappearing, and thanks to AI, customer interaction will be all about the good stuff. And if we allow the technology to take us down a path that’s good for customers, it will also be good for us
Who knows? How could we know, anyway?
Our host kicked off the third DXC Zoreza Global-sponsered insurance podcast with reference to Observer journalist John Naughton’s recent article about this very topic.
The writer asserts that an investment bubble goes through five stages: displacement, boom, euphoria, profit-taking, and panic. Apparently, we’re experiencing euphoria at the moment, but it’s the next stage, profit-taking, which concerned me and fellow podcast panelist Tim Yorke.
“Since nobody is making real money yet from AI except those who build the hardware, there are precious few profits to take.” “This generative AI turns out to be great at spending money, but not at producing returns on investment,” writes Naughton.
Previous podcast guests have sounded similar cautionary notes on AI, reminding listeners that real-world return on investment is hard to achieve, particularly in a regulated industry.
So, what does it mean for the insurance industry?
Tim, Group Transformation Director of Ecclesiastical Insurance (owned by The Benefact Group), claimed his company was cautiously optimistic.
“If this were 3 or 4 years ago, we'd be talking about blockchain. These things go in cycles, but I think there's something quite different about AI. For instance, what exactly do we mean by AI? It has many segments, and we need to know we’re using the right tools to tackle the right kind of problems.”
He pointed out that we have no problem with machine learning. It’s fairly standard, high-powered computing that allows us to respond to predictable circumstances, making better decisions by pushing more data into a process we understand.
That’s why Ecclesiastical is cautiously optimistic. Insurance companies do a lot of data crunching and routine processing, so there's certainly an opportunity to generate value. Beyond that, the most immediate benefits are in decision support, supporting underwriters as they process masses of data and looking for those crucial patterns that inform the best decisions.
We’re not using “AI,” we‘re using “I”
Tim said that as Ecclesiastical sees data as a mostly analog process, “Machine-learning-type applications would help us a lot. We’d see more patterns and find more opportunities, but as we all know, correlation isn't causation. If that were the case, there would be a dogfight to insure the risks that were demonstrably worth insuring. And when no one was making money on those preferred risks, there’d be a battle for the rest of an uneconomical pie.”
One thought caught my imagination. He asked if either of us had opened a bank account recently because the traditional six-week process has been reduced to minutes, a remarkable development driven by technology.
I don't think we've quite captured that within insurance yet. However, although AI might not take insurers to that kind of business pinnacle, it will definitely enable insurers to make fact-based decisions supported by quality data and, in turn, make them more profitable.
Rationalizing expert guesswork
Thankfully, the old managing editor’s maxim, “Never let the facts get in the way of a good story,” no longer applies here. We can use AI to substantiate an experienced professional’s gut feeling via an astonishing amount of data and advanced analytics, to fuel optimal decisions and promote a more empirical truth.
The conversation turned to employees and how resource management has evolved. Insurance companies used to have departments full of “wording” technicians, for example. Now, they’ve been released from their airless basements, and we can use much of that embedded data, knowledge, and experience to complete tasks like consumer-duty wordings at a faster rate, at a lower cost and within machine learning parameters.
It’s a sign of the times
My two daughters, 15 and 17 years old, are constantly on the phone, but they refuse to take a call from anyone they don't know. The girls would rather communicate via WhatsApp, Snapchat or some other flavor-of-the-month app. This is a challenge for insurers in the future because, to remain relevant, they must add communication channels favored by the younger generation to their capabilities.
“Look at things like policy documents,” said Tim. His children are young adults and they’re buying insurance for the first time. To them, it’s an unbelievably archaic process. Even if they get a sign-in link to a vault, they’re faced with a 20-page cover summary. Insurers still haven't worked out how to connect products to the needs of future customers who are unused to doing things the same way as their parents. However, AI will enable insurers to move with the times. The Ecclesiastical team has been exploring how to communicate vital information as quickly and simply as possible. They believe that people shouldn’t have to be paralegals to buy insurance.
Change partnerships
“We're an insurance company, right?” challenged Tim. “We know what we do, what we've done, and the technology we've got. But with all the stuff that's emerging really fast, keeping up to date and engaged with everything is a massive challenge. Just banging in a point solution here or there is incredibly difficult for integration.” He believes that the nature of technology provider partnerships must change. Insurers must engage with specialists who can guide and support them on the digital transformation journey.
This is fine, but I think the days of betting the farm on replacing a massive core system with something that's hopefully going to completely transform every process, are more or less behind us. Timelines are shrinking, so we need to be able to move faster, respond quicker, and integrate with external partners while operating in both new and traditional worlds.
The key to future success is to allow your core system sufficient flexibility to bolt on apps and services that do different jobs, making small wins and interacting with a core solution architected to take advantage of emerging developments.
Keep your eyes on the prize
Profitability is about evaluating the size of the prize. When you make a move, what are the benefits? What’s the return on your investment? Now, that return might be more strategic than financial, but every move must align with your business goals.
DXC Zoreza Global’s initial approach is to understand what the most significant elements of a problem are. We feel it’s imperative to involve internal analytical teams when discussing the use of AI. Insurers are quite protective of their IP, and we don't want to detract from what existing teams are doing but embrace it and work collaboratively. We support the existing business, making sure we’re in sync and completing short engagements where we can deliver one or more benefits at each stage.
We’re drowning in data
DXC Zoreza Global does a lot of work with the automotive industry and vehicles, where onboard computers collect masses of data traditionally held in a black box. It’s said (although I don’t know by whom) that an autonomous vehicle produces 19 terabytes of data every hour. But only 10%-15% is quality data. So, you have to use AI and machine learning to work out how to extract the good stuff.
And that’s a real challenge
One of the challenges of preventative maintenance and so on is that the sheer volume of data is difficult to transmit and use in real-time. So, is having too much data the actual industry challenge that needs to be addressed? Well, maybe, maybe not. Unless the black box was added by an insurance company, a lot of driver telematics data is held by the car maker. You don’t get much usable information from it, but you can see why so many insurers are desperate to partner with motor manufacturers.
We talk glibly about the importance of collecting and using data in the future. The fact is, we're producing masses of data already through AI and machine learning. Deriving the benefits of that data in a quality way that supports the business and future activities is the hard part.
Move forward in a carefully structured manner
So, what will significantly affect how we use AI and machine learning in the coming years?
How about turning unstructured data into structured data? Processing a paper invoice, say, is a task that would traditionally take a person forever. Perhaps it’s in a format they haven’t used before, so they type it in manually. Instead, machine learning could reference the OCR (optical character reference) and make predictions, pulling out key informational clips from those images. That's a real, tangible benefit.
“However, there's a danger,” warned Tim.” If we all ask the machine the same question, we’ll get the same answer. Where do we go from there?”
Look before you leap
We have to understand how to curate the data sets to drive these machine-learning-type applications. This is one of the new aptitudes we don't have right now. It's the R&D we've been carrying out for some time. It's about new capabilities and new skills, and we need to figure out how to develop them. That means working with partners who can help us see the big picture.
Of course, it's not cheap. Companies move to the cloud and think it's a great leap forward. Then they consider the compute costs — the more data they push through, the more expensive it gets. So, we must be crystal clear about the value generated.
One door opens, another slams shut
In a commercial world, clients have to show an equitable return. How does DXC Zoreza Global help them wade through all the lumber to work out what they should be doing? As if we were considering a chess move, how does that fit in the context of everything else that they could be doing?
Whichever move you make shuts off a whole load of other possible moves.
It’s all about context
This leads straight back to the architecture question because understanding the large-scale context into which they want to deploy rules out a bunch of other activities. For me, it’s a bit like when I went through a phase of thinking conferences were dead in the water — just the same old people talking about the same old things, and what do you really get from attending anyway?
Now I think, great, a conference. A fabulous opportunity to engage with people who explore possibilities in many different ways. Lots of vendors under the same roof, talking about new ideas they're experimenting with. And that's brilliant. I encourage my team to attend whenever possible because it's part of the all-important thinking process that generates fresh ideas and, ultimately, profits.
What about in 5 years’ time?
Historically, most customer-insurer interactions have been prompted by the need to collect data. Probably 30%-40% of business costs are related to data collection. Tim said that when he started, 35 years ago, work was all about filling in forms and using the postal service. Obviously, the internet shifted a lot of that activity onto the customer.
“Some things don’t change, however,” he added. “I don’t know if you’ve tried to get multiple insurance quotes, but it's a royal pain. Even using the comparison sites, you often have to input the same tranche of information each time.” Tim felt confident, though. “AI presents us with the chance to transform that process. Punch your postcode into the machine, and it will tell you pretty much everything we need to know. I mean, we can do that for flood risk, we can do it for substance risk, we can do that for all kinds of things.”
More meaningful customer interaction
Compiling non-competitive, public-domain data is nothing like the old tedious interactions. In particular, claims handling is changing massively. A lot of the data collection grind is disappearing, and customer interaction is reserved for when we need critical decisions about the extent to which we're going to make payouts, offer risk protection or provide support for other services. “Customer interaction will be all about the good stuff. And if we allow the technology to take us down a path that’s good for customers, it will also be good for us.”
Data will be better quality — decisions, too —and AI will enable insurers to compete in markets where they can differentiate themselves. It’ll be about their ability to make the right decision for customers, as well as themselves, based on an increased volume of AI-driven risk factors and a robust, value-generating proposition.
So, is AI in insurance a dot.com-type bubble waiting to burst?
Who can say? But it doesn’t stop you taking advantage of the benefits of AI and automation to steal a march on your competitors in the meantime.
Does it?
Join the conversation
If you’d like to discover how DXC Zoreza Global can make AI, machine learning and automation work for your insurance company, visit our website or contact us.
Simon Hunter
Simon is a highly experienced Product Owner and Technical Architect with a strong track record in the insurance industry, particularly in delivering advanced claims solutions. With a solid foundation as a Java developer, he excels at understanding complex technical challenges and providing structured, effective solutions.
