Generative AI is one of the most talked about technologies of recent times – and that’s a competitive field. The rise of ChatGPT was quickly followed by apocalyptic headlines about superhuman artificial intelligence posing a real and present threat to humanity, particularly when its makers at OpenAI called for an international regulatory body equivalent to the world’s nuclear watchdog to help manage the “existential risk” it poses1.
Already, however, some of the hype is being tempered by reality. There is no doubt the technology is astonishing, capable of rapidly reading and synthesising texts, scouring the internet for data and composing emails, fixing coding errors and automating processes at lightning speed. Spend some time engaging with ChatGPT, and its lookalikes rolled out in a Big Tech AI arms race, and some of the technology’s limitations quickly become apparent, whether it’s the capacity to “hallucinate” seemingly plausible yet entirely fictitious facts or its vulnerability to reproduce and reinforce bias.
It is, of course, entirely dependent on the data it is fed, and does not yet have the knowhow or “common sense” to make
logical deductions from the data it accesses. When RGA, a global reinsurance company with a focus on life- and health-related reinsurance solutions, for example, put the technology to the test, they found it performed admirably when asked about general life insurance topics but when presented with a specific health-related underwriting scenario, ChatGPT failed to recognize that a child cannot inherit a genetic condition from an adoptive parent2.
Is it friction or empathy?
There are real opportunities, however, to use AI to transform much of the leg work of the insurance industry, including powering chatbots, processing large amounts of data to identify patterns and anomalies that may indicate fraudulent activity and generating content such as policy summaries and coverage explanations. It’s also an enabling technology for low-touch or no-touch claims, which is considered the next frontier in customer experience.
The technology is certainly capable of virtually eliminating friction from simple low-value claims – but is it desirable? The industry is undecided.
“We already have the capability to deliver low-touch/no-touch claims. What’s more uncertain is whether the customer wants this?”
Nick Kelsall, Allianz
“We already have the technology and capability to deliver low-touch/no-touch claims,” says Nick Kelsall, Head of Insurance Claims at Allianz, which not only has a stake in Lemonade, the insurtech that famously set a world record when its AI-powered bot settled a claim for a lost cost in just 3 seconds, but also offers its own no-touch process for bent metal motors claims. “What’s more uncertain is whether the customer wants this?”
Chirag Jindal, Head of Insurance, at ServiceNow, which offers a cloud-based platform for automating digital workflows, points out that even customers who are comfortable with digital experience will often seek out a human when they file a claim.
“It’s a moment of distress when customers crave empathy and reassurance,” says Jindal. “It’s so important to get this right because claims can be the highest point of attrition.”
“There’s a lot of conversation in the industry about straight through processing of claims versus an empathetic high touch experience but in reality, a lot of people will want a balance between the two.”
The technology works then, but the customer appetite lags.
Niamh McConville, Head of Claims Operations at Generali, says this presents a challenge for insurers, who must resist the temptation to force customers down the lowest-cost-to-serve purely digital channel.
“You have to make it easy for the customer, so if they’re on the digital journey they can see the phone number and they can jump in and jump out at any stage to speak to a human and have that supported interaction,” she says.
Legacy workarounds
It’s not just customers that still want the human touch. Many carriers still have legacy systems that require it.
“Many companies have a technology stack that is still very archaic and still can’t provide real time information to brokers, agents and customers,” says Chirag Jindal of ServiceNow. “It’s going to take a few phases to clean up the tech stack so it’s ready to consume the outputs of generative AI.”
This isn’t the first transformational technology that has seen its disruptive potential, at least temporarily, derailed by the hard reality of the industry’s toxic legacy debt.
“A few years ago, everyone was talking about the potential of blockchain and smart contracts to transform insurance; months later no-one talks about it,” observes Jindal. “It’s not because there’s anything wrong with blockchain – the insurers’ systems just aren’t ready for it. And the big issue with generative AI is how to combine its outputs with enterprise data. That will be critical, and insurers aren’t there yet.”
“ChatGPT is more advanced than most insurers are ready for”
Niamh McConville, Generali
“ChatGPT is more advanced than most insurers are ready for,” agrees Niamh McConville of Generali.
“Insurers recognise that what they need is to put in some system of orchestration so they can manage the dichotomy of legacy stack and the modern API economy”
Chirag Jindal, ServiceNow
He points out that most insurers have now had the realisation that those legacy systems aren’t going anywhere. “They now recognise that what they need is to put in some system of orchestration so they can manage the dichotomy of legacy stack and the modern API economy. There’s no point using a drone-imaging insurtech for real-time loss analysis if it can’t talk to the core.”
Putting in the orchestration to bridge between the two worlds is now a priority for insurers. “One customer said to me that ‘we didn’t start out to build an inefficient process, we did a lot of workarounds to get here’,” says Jindal. “And that’s where we are now. We need smarter workarounds.”
This is certainly echoed by Nick Kelsall at Allianz. He stresses the importance of rethinking processes, so they are truly focused on the customer journey. “Work out how you can optimise processes, what questions do you really need to ask, what information do you already hold, and what does the customer really want? Look at the process first before you buy in shiny new tech, or we will just keep putting our people and our customers through this madness.”
This backroom re-engineering work could prove more valuable than the frontline chatbot.
“It may not be as visible to the customer because the experience of orchestrating sits behind the scenes,” said Jindal. “But it could really transform operations.”
Indeed, Nick Kelsall of Allianz believes many of the real wins of the AI tools will be internal, optimising processes, improving efficiency and even supporting the ESG agenda. “It can be a very powerful tool for sustainability,” says Kelsall. “Supporting repair over replace, sourcing parts for vehicles and stopping customers having to drive 200 miles to get their vehicle fixed. It’s good for business, customers and society.”
Can empathy be automated?
AI that can speak to a customer in a human-like way, detecting complex emotions such as frustration or anxiety or sarcasm, could create digital journeys that blend convenience with empathy.
“Empathetic AI is the next step but it’s going to take time in insurance”
Niamh McConville, Generali
“Empathetic AI is the next step but it’s going to take time in insurance,” says McConville of Generali. “When you look at an AI like Siri or Alexa, they are taking in huge amounts of data and learning from it every day, but they still have issues with emotion and nuance. An AI tool for insurance claims is going to have far less data coming into it so we will not learn as fast. It will take years to become empathetic and it still will not be true to us as humans.”
And there are some big questions to be answered around
governance, data and ethics.
“Are we truly empathetic if we are forcing our customer into the process of talking to a machine and are we really ethical if we are masking that they’re taking to a machine?” ponders Nick Kelsall of Allianz. “Most people would answer no to those questions.”
For now, the focus is on automating the simple manual tasks and speed straight through processing of low-risk, low-value claims. Further ahead, however, here’s the potential to augment analytics to automate more complex claims.
“Parametric insurance is tailor-made for AI”
Niamh McConville, Generali
This is already underway, with data sets from drones, remote sensors and connected devices helping to power advanced algorithms that can deliver on-the-spot assessments of damage, assess claims and spot fraud.
“We can use drones for 360 scans of large property claims from day one to augment analytics,” says Nick Kelsall of Allianz while Niamh McConville of Generali highlights the rise of parametric insurance in agriculture and property claims. “It’s tailor made for
AI,” she says.
For some complex claims, however, the pain points are not analytics gaps but the very structure of the system – and fixing this is beyond the remit of any one insurer.
“In complex claims you typically have an opponent and it’s this adversarial set up that typically frustrates the process,” says Nick Kelsall of Allianz. “With the loss of an arm, for example, we can accurately estimate what the loss is on day one, but it might take the victim two years to receive the compensation because it’s down to the lawyer on the other side and as long as those parties have skin in the game then it’s very hard to change that. We really need a culture shift on this for the benefit of victims.”
Ten years ahead
Culture shifts come slow, but AI is advancing at lightning speed. In ten years’ time, what will the balance between human and digital look like within a claims department?
“We’re already recruiting differently – now we want data literacy, problem solving and empathy because the technology can already supply the claims technical support.”
Nick Kelsall, Allianz
There will still be humans, but their skill sets may look very different, says Nick Kelsall of Allianz. “We’re already recruiting differently for claims handlers,” he says. “You don’t need to a claims technical expert – we’re looking for different skills, we want data literacy, problem solving and empathetic skills because the technology can already supply the technical support.”
Niamh McConville notes that ten years is a difficult time frame. “A lot can change, or nothing can,” she says. “But there will still be humans, using their empathy to deal with customers, while AI bots will sit within claims with the insurance and technology knowledge while technical claims handlers will be there to handle the more complex claims and develop AI tools.”
“We’re going to see augmented human intelligence, providing proactive support, guidance and empathy”
Chirag Jindal, ServiceNow
Chirag Jindal of ServiceNow agrees. “We’re going to see augmented human intelligence, providing proactive support, guidance and empathy,” he says.
Next-generation AI may be in the background, automating manual tasks, surfacing data and spotting risks, but the frontline, for some customers, some claims, some of the time: it’s still very human.
Chirag Jindal
Head of Insurance, Americas, ServiceNow
Chirag brings over 12 years of experience of serving Property & Casualty and Life & Annuities insurance carriers. He has experience in solving complex business issues including Operating Model Design, Agent and Customer Experience, Distribution Strategy, Underwriting Operations and Claims Automation. At
ServiceNow, he is responsible for the Go-to-Market Strategy in Insurance – use case activation and Partner / ISV strategy in Insurance. He is focused on solving complex business problems while elevating the Customer, Employee and the Distributor Experience through the ServiceNow platform.
Prior to joining ServiceNow, Chirag was an Associate Partner at Deloitte Consulting where he led several Digital Transformation for top Insurance carriers in the United States. He has spent time in solving business led tech enabled programs to focus on key issues in the industry such as – 4th year agent retention in Life Insurance, Submission Challenges in Small and Large Commercial, Claims Automation, Service Transformations.
Chirag is originally from Delhi, India and now lives in Chicago with his wife and 2 girls – 6- and 2-year-old. He completed his undergrad in Computer Science from India and MBA from Kellogg School of Management, Northwestern University.
Niamh McConville
Head of Claims Operations and Technical Support and Parametric Claims, Generali
Niamh McConville has a degree in Psychology. She has worked in many areas of Insurance, including Life, Municipal and High Net Worth. Niamh joined Generali in 2022 as the Head of Claims Operations, Technical Support and Parametric Claims.
Nick Kelsall
Head of Insurance Claims, Allianz
Nick joined Allianz in the UK in 2015 as technical claims manager, progressing to fraud manager and head of insurance claims, with a secondment to head of motor. In 2020, he moved to Allianz Group in Munich. As senior global P&C executive, he led on the creation and implementation of a globalised digital fraud solution. He then became part of the Allianz Customer Model (ACM) leadership team with a focus on improving claims processes and aligning supply chain networks globally. After this experience developing IT assets that support the claims journey, he joined the UK technical claims
team as head of motor.
In partnership with:
This article is sponsored by Servicenow
ServiceNow makes the world of insurance work better for everyone. Over 400 insurance companies globally use ServiceNow’s cloud-based platform and solutions to intelligently automate processes and services across their enterprise. So insurance companies can modernise their employee engagements, risk management, security, and technology innovation. And we can all create the future of insurance that we imagine.
Learn more: https://your.servicenow.com/uki-insurance
Stay current with Insurance Innovators
Subscribe to our newsletter to receive news, insights and special offers.