Why insurance is quickly becoming the ‘I’ in AI

Popular News Tags (5289)

  • There’s a growing sense in the technology industry that we are on the cusp of an era in which the technologies that previously existed only in sci-fi storylines will become increasingly ubiquitous parts of the fabric of everyday life.

    Artificial Intelligence (the technologists’ favourite) in particular, is beginning to reach a level of maturity that means it can be deployed in businesses – a major step in the evolution of a technology towards mass adoption.

    Much of the AI deployment that took place five years ago was very specialist, and had relatively few capabilities. Since then, improvements in processing power have allowed huge developments in the kinds of tasks in which AI can be deployed.

    More than that, the approach that IT professionals are now taking to the actual implementation of machine learning has made it a far more effective tool for delving into problems of an increasingly complex and diverse nature.

    In Madrid, we’re already working with healthcare consultants to help them make accurate diagnoses in just a fraction of the amount of time it would previously have taken. Machine learning technology allows broad data sets of clinical trials, treatments, studies and anonymised data about previous patients to be parsed within seconds – freeing up time for the consultants to put their brains and expertise to use with their patients.

    Healthy new applications

    We’re able to transfer much of what we’ve learned through this work in the healthcare space to other sectors.

    Insurance – an area not far removed from healthcare in some cases – provides plenty of opportunities for overlap. From actuarial calculation of risk or claims processing, to understanding of risk, there’s plenty of room for AI to be used to improve services and free up human agent time.

    What’s most exciting is that we are beginning to be able to provide technology to support some of these more complex roles.

    Previously, AI was there for taking on simple, largely administrative tasks – remember Microsoft Word’s much-maligned Clippy? We’re now getting to the stage at which more subtle thinking can be completed by the computers in our charge.

    It’s still relatively specialist – you couldn’t take the healthcare AI system and apply it to fixing a car without a major overhaul – but we are able to provide solutions for specific problems that evolve over time. Through a process of machine learning, AI systems are developing their knowledge and aptitude alongside the problems they’re designed to solve – and therefore developing in efficiency over time too.

    Data for good

    The finance sector, and the day-to-day of insurance brokering in particular, generates a huge amount of data that’s used to assess and respond to risk. We’re able to use these data sets in combination with AI programmes to segment sectors within the insurance market and come up with risk profiles more quickly and effectively.

    It also opens up the ability to provide greater personalisation of insurance products – something that’s key in an increasingly customer-focused industry. In a recent Experian report into personalisation, 79.3% of respondents stated they considered it important – more than half (51.7%) said it was very important to their organisations.

    Quicker and sharper – and cheaper

    The potential benefits are myriad.

    Cost is an obvious one – if you can reduce the number of people required to complete a task, such as analysing photos submitted with a car insurance claim, then that frees up time and human resources for other things. And time, of course, is money.

    Related to this is speed – another key benefit. Whether providing a diagnosis in the aforementioned healthcare scenario or processing an important insurance claim, time and being able to use it efficiently is ever-critical.

    Often, however, speed comes at the expense of accuracy. Thankfully, in the case of AI, this is not so.

    One thing about machines is that, although they do make mistakes, they tend to get better with training and once they are traineddon’t tend to forget things. Humans, on the other hand, aren’t always so blessed. It’s hardly a coincidence that most of us now use sat navs to help us find our way around  alleviating the need to remember even the routes we take most days.

    Changing the face of insurance

    These developments couldn’t have come at a better time for the insurance industry. It’s a sector feeling the pinch from aggregators driving down price (while risk and pay-outs remain the same), and often finds itself battling with reputational issues – many of which revolve around the argument that profits are prioritised over customer satisfaction. And this is all before you consider the serious impact of fraudulent claims.

    The focus of this is so often on price, with competing brokers feeling pushed to offer cost-conscious customers simply the cheapest service rather than being able to concentrate on a more rounded ‘value for money’ approach.

    AI is helping to change this.

    Cost is of course important – and company savings can be passed on to customers – but more than that, adopting AI-supported processes is allowing insurers to provide more timely and personalised customer service, resulting in improved customer retention levels.

    Who wouldn’t want an insurance company that can resolve issues quickly and easily, as well as offering an attractive price? It’s perhaps the most effective way to endear yourself to a customer, and AI is helping make the effective simple too.

    There are plenty of areas of insurance open to the implementation of AI, right the way from segmentation of markets and ways to aim products, to developing new services, on boarding new customers and making that process simple and easy, or opening up different customer service channels using natural language processing to allow people to go through the claims process with a virtual agent.

    Rewards for risks

    This is not a strategy that is entirely risk or challenge-free, of course.

    There’s quite a lot of hype in the market at the moment, but it’s understandable given the potential benefits. What’s most important at this stage is to make sure that the people responsible for running AI programmes really know what they’re doing.

    You need to know what you really want to get out of your AI implementation –  the technology itself is no longer the complicated bit; it’s how you use it that really matters, particularly given the specificity of AI applications.

    Doing it wrong or ineffectively could end up costing you a lot of time and money. The solutions are there, but it’s about having the knowledge to put them in and make them work for you.

    Ultimately, though, when it comes to risk and the question of whether or not to start implementing AI-led strategies, I’d say the main risk is in not doing it. Because regardless of what you do, you can guarantee your competitors will be using AI to improve their service in the future – if they’re not already doing so, that is…

Share this story