Artificial Intelligence (AI) is at the heart of everything we do at Nuon. So, who better to talk to about the opportunities it presents for the insurance industry than renowned AI expert and Nuon advisor, Professor Andy Pardoe.
We sat down to talk to Professor Pardoe about why consumer forces are directing AI development, how AI is changing insurance for the better and why insurers shouldn’t wait to get involved.
Professor Andy Pardoe is a world leader on the subject of Artificial Intelligence and is in constant demand as a consultant across industry sectors, including financial services, retail, media and, of course, insurance.
An expert advisor to the British Government’s All Party Parliamentary Group on AI, Professor Pardoe is also listed by IBM Watson as one of the Top 30 AI Influencers globally. He holds three degrees from Warwick University (BEng, PhD, MBA), an Honorary Professorship and is Chair of the Warwick Technology Professional Network.
Professor Pardoe, thanks very much for chatting with us today! Let’s break things in gently – for those who are coming at this cold, what exactly do we mean when we talk about AI?
Artificial Intelligence, or AI, is software that can perform tasks that, previously, were considered only possible using human intelligence. When Google, for instance, looks at the first few letters of your search, their software draws on huge volumes of data and then uses a smart algorithm to predict what you’re about to type. That’s just one example of AI that is already part of our everyday lives.
AI seems to be working its way into every area of our lives. How is it impacting the insurance industry?
From the outside in. The insurance industry is hundreds of years old and sometimes it feels like it. The industry is trying to modernise, and progress seems slow right now, but it’s speeding up because of external pressures.
Take the motor industry, for instance. We’re still some way off the mass adoption of totally autonomous self-driving cars, but we are seeing modern cars with the ability to collect a wide range of data points, either through the car’s software or by connecting to smartphone applications. There are endless possible applications for applying AI to this data, but one method related to the insurance industry is to give drivers a premium based on their driving style. There are already at least half a dozen different companies in the UK – DriveScore and Rooster, for example – that will offer you personalised premiums based on how you drive, where you drive and when you drive.
A 21-year-old driver who’s only held a licence for six months would traditionally be considered a high risk by the insurance industries and would be charged a premium to match. But he or she could use one of these apps on their phone to track their driving. If they only drive a couple of times a week during low-risk times of the day, brake smoothly and accelerate gently, they could be offered a much lower premium than if they were being assessed purely on age and experience.
This has the potential to completely upend the motor insurance industry.
Another interesting case, although not really related to AI, is the car subscription model. This is different to leasing, partly because it requires shorter commitments and lower deposits, but mainly because it’s an all-inclusive program that includes all related motor costs except for fuel. Critically this typically includes insurance, completely cutting the consumer out of the transaction.
I actually have a car subscription myself and it’s a very interesting model that, again, has the potential to reshape the industry.
So, these are examples of new technology products that affect the insurance industry, or have the potential to affect it?
Exactly. And what these two examples have in common is that they’re both offering a personalised service that smartly taps into what consumers are looking for.
These days, consumers are much more savvy about their data, particularly in terms of recognising how this can provide a much more user-friendly, individualised service. Instead of being offered just 2-3 options based on their most basic demographics, they can provide exponentially higher levels of data and as a result be offered something designed just for them. Now that consumers know this is possible, they’re increasingly going to start expecting it.
This is what I mean by AI shaping the insurance industry from the outside in. Innovations happening outside of the industry are creating opportunities to develop AI-powered insurance products, or could even be said to be putting pressure on the industry to adjust.
If, for instance, the car subscription model becomes the norm, motor insurers are going to have to find a way to respond to that.
Do you see a future in which consumers demand this level of personalisation, and may even reject insurers that don’t offer this approach?
I think it’s inevitable. Consumers already know that they’re paying higher premiums to offset higher risk drivers in the same demographics. Once they become aware that they don’t have to simply put up with this, and that they can be insured based on the fact that they mostly make short, local car trips, and never exceed the speed limit, insurers will have to start offering this option or they’ll be left behind.
And that’s only in the short-term. As personalised premiums become the norm, the natural next step is to allow drivers to see how their driving choices are affecting their premiums in real time. Imagine putting your destination into your satnav, and being given the choice of a route that is ten minutes quicker, but will add an extra £1.50 to your insurance payment this month. Now the consumer can choose between taking a slightly longer route, or paying a little extra to get to their destination faster. Or imagine your insurance app suddenly announcing that you’ve improved the smoothness of your braking by 10% over the last month and, as a reward, your premium is going to be 5% cheaper next month.
Again, notice that this is all about giving consumers choice and allowing them to feel as if the data they’re sharing with the insurer is being rewarded financially. That’s the direction we’re moving, and that’s the direction in which insurers should be thinking.
Many insurance companies are already starting to implement AI into their models. Is it the right time to move on this, or if you haven’t yet started is it already too late?
I don’t think it’s too late. But it’s probably something you shouldn’t put off for too long.
When an industry starts to see AI working its way in, you always get both extremes. The early innovators lead and have the potential to win big. While others wait too long and end up paying a premium to catch up with the rest.
If you have the resources and the ambition to innovate, and you can be bold, there is huge potential to win over a share of the market. If you’re more risk-averse, the best approach is to at least make preparations now. Explore the options to introduce AI into your business model, test it carefully, and then apply it once it makes financial sense to do so.
We’ve been lucky enough to work closely with you on our own AI-powered insurance software. How does Nuon’s approach fit in with what we’ve discussed today?
Nuon’s approach is interesting because it takes what insurers are already doing, and uses AI to allow them to do it faster and with more accuracy.
For example, an insurer who promotes their products on an aggregator, such as GoCompare, can use Nuon to rapidly adjust their prices in response to unexpected movements in the market. The result is that they can either regain market share when it dips, or they can test higher prices to make sure they’re not leaving money on the table.
It’s innovative, but it’s based on a proven approach to AI, so it doesn’t represent the kind of risk that companies who have developed driving scoring apps are taking. Nuon is essentially allowing insurers to improve the results of what they’re already doing, rather than completely upend their products.
It also fits in well with what we just discussed about preparing and testing. Nuon’s products can be trialled through a sandbox so you can see how the AI would work in practice, without taking any real risk. You can then apply it in the real world when, and only when, it makes financial sense to do so.