AI is a term that applies to a variety of different technologies that deploy computer ‘intelligence’ to some degree.
The first level is robotics, which is really a flash name for process automation. With this, the computer follows a set of rules that you’ve written, so it can be a good way to sort junk email or populate an application form from existing information.
Machine learning takes this a step further, training the computer to carry out a task. As an example, what’s the difference between a bus and coach? As a person, you’d know instinctively, but you can also train a computer to identify each type of transport by showing it lots of photographs of buses and coaches. Over time, it will ‘learn’ the difference, with accuracy increasing the more ‘training’ it has.
It can also be used for more complex tasks such as interpreting words and speech, as is the case with chatbots and voice assistants such as Siri and Alexa.
However it’s used, it’s not strictly the same as learning. Computers are very binary and the training process simply converts information into computer code that is then used in its validation process.
AI is being used for all sorts of process and customer experience improvements across the insurance industry. Robotics is automating back office processes and more and more firms are introducing chatbots to answer customer questions.
At Allianz we’re embracing this technology throughout our business. As an example, in January we launched our defendant hub, a digital proposition that uses AI to enable our claims handlers to action Ministry of Justice Stage 3 claims quickly and more efficiently.
Rather than spend time checking court proceedings packs and manually downloading documents, details of the award are captured by counsel and automatically fed into our system. Doing this creates an average saving of 30 minutes for each claim.
Future applications for AI are also being explored, including its use in more complex processes such as assessing photographs of vehicle damage or interpreting policy clauses and cover terminology.
Working in a highly regulated market is actually an advantage when it comes to implementing AI.
Insurers always have to be able to explain the decision-making process they use. Failure to do this isn’t acceptable to the customer, to us, or to the regulator, so we have to make sure we understand exactly what the technology is doing before we implement it. Taking this approach enhances the customer experience, which also benefits the insurance industry. In many ways, it’s the perfect home for AI: as well as having lots of processes that can be automated, it’s a cautious industry so any changes are well tested before they go live.
The significance of the decision it’s making also plays into the level of training required: you’d expect a machine being used to diagnose cancer to be put through much more rigorous training than one being trained to identify wheels in photographs.
Insurers also need to be mindful that any new technologies are in line with customer need. Take chatbots as an example. Some people prefer to speak to a person and therefore you need to offer a variety of options to suit your customers.
In time, as the technology gets better and people become more familiar with it, adoption will increase. For instance, we all trust our satnavs now but it wasn’t that long ago that we’d check on a map whether it was taking us to the right place. The same will happen with chatbots, but for now we need to be aware of different customer needs.
Rather than think about what AI is available and how it can be slotted into your business model, think about what you want to improve or change and whether AI can help.
Don’t be afraid to start small either. AI is usually scaleable so you could start out seeing if it works for an address change, for instance, before adding in further functionality such as phone number or name changes.
And if it doesn’t work, be prepared to throw it away and try something different. This technology can sound futuristic, so it can be tempting to think it requires a different approach, but it’s just common-sense business practice really.