"If you are not using AI you will be replaced by a company that is using AI"

Survey respondent

Some interviewees share their long-standing experience with AI, dating back nearly a decade in some cases. Others are newer, ‘digital first’ brands, for whom unsurprisingly AI and Generative AI technologies are front and centre. Others again, highlight current pilot projects or areas of experimentation.

Aviva was one of the earliest adopters of AI in the industry, starting some 10 years ago. But its Claims Chief Operating Officer, Waqar Ahmed, cautions against investing in technology for technology’s sake, warning against the allure of "shiny new toys". He emphasises the importance of "discipline" in applying the right tool to do the right job, always having the end goal in sight, and improving the customer experience, "There’s no point in using an expensive tool (when ultimately the cost will be passed on to customers), where an inexpensive one could do the job just as well."

"Across the insurance industry, pilots and trials are being run to see what Generative AI has to offer in improving how claims are managed, how decisions are made, and how customers experience the claims journey. At AXA, the integration of both traditional and Generative AI is a clear example of how insurers can evolve strategically while keeping the human touch central."

The Generative AI Opportunity

Building on our traditional machine learning capabilities, we are now exploring how Generative AI can add further value, particularly in document analysis. For example, Generative AI can read and interpret medical reports or other claim-related documents to help detect early signs of complexity in a case, so it can be escalated up to our specialist team more speedily.   

"Our handlers now say AI gives them more time to focus on the parts of the job that require their expertise and human skill-sets, with much of the 'boring' admin taken away, so they have more time to focus on handling complex conversations, providing empathetic support and making decisions. All aspects of the job that really benefit from the human touch."

Waqar Ahmed that agrees success comes from involving frontline colleagues in the design of customer-facing technology. This works not just in terms of the best way to ensure colleagues buy-in to the use of new technologies, but also from the point of view of making sure customers’ needs are the focus of any tech project. He describes how positively colleagues in his claims teams have embraced new technologies and Aviva’s mission to become "a data-driven intelligence-led function, where people are at the heart of what we're doing". 

“There is often an assumption that when you deploy AI, staff perceive it as a threat”, he says. “But we have proven that when it works side by side with claims handlers to augment what they do and assist them in their roles, they are then able to amplify their own human capabilities, such as empathy, which is so crucial to the customer experience. We have seen their engagement with the company increase.” 

Further, Andrew Wilkinson of the Motor Insurers' Bureau believes AI can act as a highly effective trainer for claims handlers, helping people learn about the relevant case law and legal complexities on the job. “DACB’s excellent AI tool for credit hire is a great example of AI at its best”, he says. “Whereas previously a handler's learning was in large part by trial and error over time, this tool helps them give an offer and explains the rationale, so helping them as a virtual colleague/assistant best friend whilst training them at the same time."

He is very upbeat about the opportunities AI presents for his colleagues, “We see the potential for AI in our processes as very positive. We don’t see it as placing jobs for the humans under threat, but instead increasing opportunity for our people, making their jobs more skilled and interesting.”

Ian Kershaw, Vice President of Customer Service, Claims and Fraud at Zego, agrees, “I don't see AI necessarily replacing claims handlers. It will just change their roles. So instead of handling 200 claims each, they’ll proactively be able to handle 800 because they'll be supported by technology that enables them to work faster and do more, and potentially to an even better standard as more parts of the process become automated.”

AXA Insurance’s Alexandra Price says it’s about taking a collaborative, transparent approach. At AXA, she tells us, “training and support are embedded into AI rollouts, and feedback loops ensure ongoing refinement. And crucially, AI projects are developed with the business teams, not handed down to them.” 

When considering the appropriate level and type of technology investment for our organisation and what Generative AI has to offer, the key for us, as with all advancements in technology, is not just to look at the benefit it will deliver for NHS Resolution, but also the wider system we operate in. We are in a slightly unique space because we operate within the health system and are also juxtaposed with the justice system. We are looking for advancements that will give us greater visibility of the ‘concerned’ space, i.e., help us identify where something has gone wrong to a significant degree, which we can then investigate in greater detail and in a wider context, giving us valuable insights to share with other parts of the healthcare system. So, when it comes to deciding on the right technology, for us it is not just about considering what gains we can make for ourselves operationally (saving operating costs, improving consistency and fairness in our decision-making, with less resource), but more about whether or not it will benefit us in what we can deliver back to the health service, both in relation to policy development and also in respect of safer clinical care. But that’s maybe where we come in with a unique perspective, because we are not profit making. 

 Where we are right now is in providing a more efficient system with the integration of some aspects of ML, but this in its infancy. Where we are moving to is the integration of true AI, to see what we can produce from our data that will help the wider system learn from what we see. We are currently updating our IT architecture so we can integrate AI for the benefits of both our internal processes and to provide external insights.

Whether integrating NLP or machine learning (I see the two as quite distinct), you need to have not just the right technology platform to support it, but the right data platform as well. From our discussions with other indemnifiers, we know that everyone is facing the same challenges around this, particularly around the issue of ‘legacy data’. Tech suppliers talk emphatically about the ‘Holy Grail’ of a complete and perfect data set, this being the only true way to be sure of reliable and consistent output, but most readily acknowledge the restrictions in data. More on this later…

Where we see the benefits of bringing true Gen AI into our systems, is in using it to learn from our past experience in order to inform our staff more accurately about potential outcomes that may occur, and help with risk management in terms of flagging where the risks to the health service may lie, so we can then work with other health partners in avoiding those risks and seeing less harm in the overall system. There are probably multiple other uses of Gen AI for the future that we see that will bring a variety of benefits. For example, assisting with our pricing models and our actuarial forecasting in relation to our long-term liabilities.

We have already launched the first iteration of our new case management system in parts of our business, and the area covering claims, the largest part of our business, is due to go live with a new case management system in the next two to three months. So this is very much the here and now for us!  This has been a couple of years in development, as you can imagine.  What it will provide us with is the platform for a true AI environment that we can actually start to operate with.

There are many challenges, like everyone else in this space, and as referenced above, our biggest concerns are around the availability of reliable data, particularly the quality of data from legacy systems. We live in the real world, so that the Holy Grail of a complete and perfect data set can only ever be an aspiration. Indeed, this is why we have seen the rise of ‘data scientists’. We use data scientists, and there is a lot they can do, but there are natural limitations because at the end of the day, they are handling historic data sets, and therefore the data does not hold the level of consistency or granularity that allows correlations to be drawn. This has the potential to become a major issue when you start to apply machine learning.

You need to have the foundation for the right data platform in place, and also the right data sets, for AI to deliver its promised benefits and produce appropriate results that are accurate and can be relied upon. Conversely, if the data is flawed, and this is what I hear quite regularly from the supplier community, then the potential is that people could build AI systems that produce results from pseudo data or from a small sample that is not necessarily representative, which then becomes a challenge when applied to a wider data set. You may then have gaps and therefore cannot produce the same or similar results to prove the output is repeatable and reliable.

"The key is to get everyone to buy into the appropriate pace of change, as well as the change itself. To do this, you need to ensure people are kept informed about the timeline and the ‘art of the realistically possible’. It’s about allowing people to understand you can only move at a certain pace and that moving at that given pace is a critical aspect, bringing people on the journey with you and dispelling myths along the way.

Of course, these characteristics are common to any change initiative. They are exacerbated when external pressures are at play, for example, the Government wanting all its agencies to invest heavily in AI at the moment.  Then people see technology transformation as something that has to be done, rather than a nice-to-have investment opportunity for the business."

"There is a lot of discussion at the moment within the wider AI space about how it is going to be regulated. I think the risk is that the legislative framework will always trail invention and innovations in the tech space. We have seen this historically. I believe the key is for an organisation to understand and set its own risk boundaries, to remain within these regulatory frameworks, and adapt as the law evolves. There are likely challenges coming down the line, and across industries, in relation to how the Data Protection Act interacts with the potential of AI and the ingestion of different data models. I’m talking here about the wider environment, not just at the organisational level but maybe even broader."

Another key element is having a clear data strategy, not just for operational efficacy but for regulatory compliance. You need to be ethical in how you go about collecting, storing and using data, and how you intend to utilise the outputs for any models you are building, whether for financial provisioning, decisions in relation to claims management, or for delivering those insights to the wider health community. Whatever those models are designed to do, you need to have a strategy in place to ensure the models themselves and the way you are using the data are appropriately assessed on a regular basis, to ensure that bias isn’t creeping in and that they are producing reliable results consistently. 

In addition, I think there will no doubt be frameworks which will be brought in to regulate how organisations can actually use AI in certain situations for decision-making processes. I think it has to be down to the individual organisation to ensure it sets its own risk appetite accordingly against those regulatory frameworks. This is a recurring conversation internally at NHS Resolution: how our risk appetite fits with the wider technological advancements that might be coming into our environment now and in the future. It’s a double-edged sword; on the one hand, we want to stay well within the regulatory confinements and ensure that the way we use technology is both ethical and on the right side of regulation. On the other hand, we are balancing our risk appetite statement with really wanting to use the new technologies positively and deriving the benefits from them. This is a difficult line for any organisation to set, especially when technology (especially in the Gen AI space) is moving at such a dramatic pace. All this needs to be reviewed continually. How regularly should organisations conduct these reviews? It is probably too simplistic to put a timeline against it, but if you did only an annual review you would soon find yourself out of date. You have to set it against what your ambitions are and what your investment strategy around your tech future looks like, and also against your operational processes, because every time you introduce new aspects of tech your operational processes are going to change, and therefore your risks may change in either direction (less/more). They may improve because you are safer, for example, advancement of some tech may make fraud detection and prevention easier and actually your risks decrease. But, on the opposite side of that same coin is the fact that by ingesting more tech-based decision making, this may present greater risks, such as biases being present. For example, if the pre-event detector mechanisms you choose to adopt end up identifying the wrong categories of individuals for fraud investigations, this could lead to reputational damage, added operational cost and most importantly, lead to a delayed claims process for genuine claimants. So, the nature of risk is going to change depending on your ingestion of tech within your organisation, and possibly improving one risk, but at the very same time heightening another. 

You can see exactly why decisions around the ingestion of Generative AI cannot be rushed!

However, Ian Kershaw describes how Zego’s experience shows generative-AI powered apps and chatbots can take automation of the customer interface to another level and achieve better customer satisfaction ratings than human interactions. "Customers still always have the option to switch to a human agent, but as we’ve developed and improved the live chat experience, we are seeing a huge reduction in the number of calls coming in to our human agents, because the chatbot is able to answer their questions completely. We see this as the proof in the pudding. Our aim is, by the end of the year, to reach 85-90% of our live chat interactions to be fully automated, by customer choice."

Another significant advantage the industry expects from Generative AI is its potential to improve the accuracy of loss assessments. As Simon Hammond points out, “There are likely many other future applications of Generative AI that could deliver a wide range of benefits - for example, enhancing our pricing models and actuarial forecasting related to long-term liabilities.” Indeed, 57% of respondents in our online survey highlighted this as a key motivation for exploring Generative AI. Additionally, 64% of respondents highlight the ability to improve consistency in decision-making as one of the main value-adds. Andrew Wilkinson says he sees AI’s ability to assist claims handlers in the way described above as one of the keys to improving consistency in the MIB’s decisions and assessing loss more accurately. 

"The model has proved so popular and successful that since the beginning of this year we have started to expand out of our commercial driver niche to provide personal car insurance too. Our focus on telematics, and pricing centred around that, means we can be particularly attractive to young drivers, also those with convictions, but who are still good drivers." 

"As a very digitally-driven brand in the marketplace, as you can imagine, AI and Gen AI are core to how we are developing our business. The biggest area we have been focusing on in terms of integrating AI is the customer service side, with the primary method our customers use to communicate with us being app-based."

We were in that very different place maybe four or five months ago. We had a very rudimentary live chat flow, like a lot of companies still have today, where the customers are just given a long list of options to choose from, to categorise their question. And, if their query doesn’t fit neatly into one of these categories, it can be immensely frustrating for them. They can get tangled up in what I call the ‘death spiral’ of just not being able to get the bot to understand their problem. They then just want to speak to a live human who can give them the answer. So, what we wanted to do was create a human-like experience on the live chat, or an even better-than-human experience, and Gen AI has enabled us to do this.  Rather than being offered a tick-box list to categorise their question, customers can now simply have a free-flowing conversation with the bot. It’s incredible how much this has improved our customer satisfaction levels. In fact, when they're dealing with the bot now, customer satisfaction is higher than when they deal with human agents!

We are really pleased with where we have got to, even in terms of tone of voice and empathy. To give an example, a customer the other day asked the question on our live chat about why his policy had been cancelled, and he mentioned he’d been out of the country and that his mother had died. The chatbot response was incredibly empathetic, ‘I'm really, really sorry to hear that. I completely understand the sort of turmoil you must be in...” 

In terms of where we place humans in the loop, of course, we monitor and quality assess chatbot interactions. They are all stored and entirely auditable. We do this in exactly the same way any insurer does with its claims handlers. Also, we don’t yet leave any decision-making to the machines. Our bot stops short of making decisions. At the moment, from a claims perspective, its role is limited to data capture. So it's taking information from the customer, that information is fed into our system automatically, but then it's a human making the decisions: is the customer at fault? Is the customer not at fault? Where does the customer need to be? What supply function does that customer need? Where does it need to be repaired?  At the moment, that's all with a human. I don't see AI necessarily replacing claims handlers. It will just change their roles. So, instead of handling 200 claims each, they’ll be able to proactively handle 800 because they'll be supported by technology that enables them to work faster and do more, and potentially to an even better standard as more parts of the process become automated.

"Do we ever envisage a future where the decisions could be fully automated? Absolutely. But, for end-to-end claims, I think we're some way off that at the moment."

"Because of this investment, we are now finding that the chatbot can answer 55-60% of customers without any human interaction. Customers still always have the option to switch to a human agent, but as we’ve developed and improved the live chat experience, we are seeing a huge reduction in the number of calls coming in to our human agents, because the chatbot is able to answer their questions completely. We see this as the proof in the pudding. Our aim is, by the end of the year to reach 85-90% of our live chat interactions to be fully automated, by customer choice."

Simon Hammond emphasises the need for humans always to remain in the loop at NHS Resolution because of the particularly sensitive nature of the claims and information dealt with: “Would we ever get to a point where the machine is telling us everything and making decisions? The risks in this are far too huge. At NHS Resolution, we are dealing with a very, very sensitive area of claims management. So, for us, we accept there will always be a requirement for an element of human decision making in all that we do.” 

Alexandra Price adds: “The traditional machine learning models we have been using for several years help claims handlers triage cases and recommend 'next best actions', but importantly, this AI is used to augment human expertise—not replace it. We have no automated decision-making within our journey. The handler is always able to review the output and accept or reject those decisions."

Andrew Wilkinson talks about a moral duty for humans not to abrogate decisions and keep responsibility for them: “We see the role of AI in assisting decision-making, but not making decisions by itself. There will always be the need for humans to take responsibility for decisions."  

Most believe that the near-term focus for AI should remain on back-office support; however, some are using it for customer-facing tasks in a controlled, hybrid way, maximising the benefits of AI-human collaboration. Alexandra Price describes AXA’s use of AI for both customer-facing and internal tasks, “In digital channels, like our Electronic Notification Of Loss platform, AI helps triage claims and suggest likely outcomes more speedily, such as whether a vehicle is repairable or whether it needs to be taken to salvage. Customers can then choose to proceed digitally or speak to a handler at any point…”  She adds: “Behind the scenes, AI models work silently, flagging potentially complex claims or summarising customer calls to save time. But the customer still interacts with a human. Ultimately, it’s the handler speaking with the customer, backed by AI-generated insights that give them more confidence in those interactions."

As Ian Kershaw explains, even digital-first brand, Zego, currently stops short of using bots to make decisions, "We don’t yet leave any decision-making to the machines. Our bot stops short of making decisions. At the moment, from a claims perspective, its role is limited to data capture. So, it's taking information from the customer, that information is fed into our system automatically, but then it's a human making the decisions: is the customer at fault? Is the customer not at fault? Where does the customer need to be? What supply function does that customer need? Where does it need to be repaired? At the moment, that's all with a human.”

Julie Plumb of Tesco’s Insurable Risks team believes particularly strongly that human oversight must remain centre stage in decision-making. “We see the role of AI in assisting decision-making, but not making decisions by itself. There will always be the need for humans to take responsibility for decisions. Experienced claims handlers, those with 20 or 30 years of experience, often have an instinct for when something doesn’t feel right, especially in cases of potential fraud or exaggeration. That kind of gut feeling is difficult, if not impossible, for machines to replicate. It’s a critical component of the process that shouldn’t be underestimated.”

However, some think this could shift in the future. Waqar Ahmed explains an alternative view: “Keeping humans in the loop at the moment is imperative, given the fledgling nature of the technology. But, I can see a time in the not-so-distant future where ‘human in the loop’ evolves to ‘human oversight'. I think this is inevitable, because if we go back to our primary focus for technology development and investment, our customers, we have the potential to build products that are ‘always on’ for them. So, irrespective of when they have a claim, they have the ability to interact with us at any time of the night or day, to interact with us so we can help them in their moment of need." But, “We must ensure that when a customer interacts with AI systems, they get the same great customer outcome that they would if they were dealing with a human handler. Wouldn’t that be a great thing for the industry? To transition that great human service to a digital experience that's always on." Ian Kershaw takes a similar stance, “Do we ever envisage a future where the decisions could be fully automated? Absolutely. But, for end-to-end claims, I think we’re some way off that at the moment.”

Andrew Wilkinson

Chief Claims Officer, MIB

“There is real value in … human-AI collaboration. We are not building AI in isolation. We are building it with our people, for our people because we believe that is how you get the best outcomes for customers, and for the business."

Ian Kershaw

Vice President of Customer Service, Claims and Fraud, Zego

“Do we ever envisage a future where the decisions could be fully automated? Absolutely! But for end-to-end claims, I think we’re some way off that at the moment."

"As I’ve said before, one of the key issues for us is ensuring that any supplier or partner we work with is aligned with our values and approach when it comes to the use of technology."

We look at claims in the context of a value chain, before a claim arises, i.e. identifying geographical hotspots for uninsured driver incidents and hit-and-runs, when an incident and a claim occurs, instructing suppliers and partners (including lawyers), how we approach negotiating settlements, how we handle data and management information, how we manage workflows and time, and how we analyse data and draw conclusions. We certainly see a role for AI in the pre-claims process, working with the police and the DVLA, for example, using cameras to predict hotspots. Also in the investigation of claims, digging to find insurers or the identity of drivers using ‘connected vehicle’ technology, such as getting information where appropriate from satnavs, phones and other internet connectivity to pinpoint who was in a car at a particular time and place, but this is several years down the line. We see the role of AI in assisting decision-making, but not making decisions by itself. There will always be the need for humans to take responsibility for decisions, but AI can do a lot to assist claims handlers. For example, collating information about the claimant’s medical records, the events surrounding the damage, presenting summaries to bring handlers up to speed more quickly and then putting documents together for experts or partners to prepare them for negotiating settlements, even perhaps using estimating assistance tools. We can see a benefit here in improving consistency in our decisions.