Generative AI is the technology no insurance company can afford to ignore. In less than three years, it has already made a huge impact – and most agree this is only the beginning of the journey. Sarah Self, Generative AI Director at Aviva, admits she’s been surprised by its rapid rollout.
‘I’ve been working in technology and transformation for over 20 years and the pace of change of the technology and the appetite for adoption is like nothing I’ve seen before,’ she says.
This is perhaps no surprise given that ChatGPT, the first of the large language models to launch, rapidly became the fastest-growing consumer application in history at that time¹, winning hundreds of millions of users within months of its November 2022 launch. People have been quick to appreciate the technology’s ability to transform their own productivity, whether it’s drafting emails, scheduling meetings or summarising reports.
'We’ve seen a 10 per cent reduction in call handling times and a better experience for both customer and agent'
In addition to personal productivity gains, many companies are seeking ways to use the technology to improve service to the customer. For insurers, one of the most impactful areas has been claims, when customers are seeking reassurance, prompt action and rapid settlement. While it may be a time least suited for the intervention of a ‘bot’, Sarah Self of Aviva says it actually ‘takes the robot out of the human’ by handling the mechanistic tasks so human call handlers can focus on what’s innately human: listening, empathy and judgement.
‘Claims summarisation is one of our favourite use cases,’ she says, highlighting how succinct AI-generated summaries not only mean handlers no longer need to put customers on hold while they catch up on the claim history but they’re also briefed on important customer data, such as vulnerability, or triggers for next steps. The result? A 10 per cent reduction in call handling times and a better experience for both customer and agent.
Claim summarisation: a favourite use case
'We’ve seen savings of four minutes a call, and with 50,000 inbound calls a month, that’s huge'
Claims summarisation has also been a powerful use case at Markerstudy Group, with Kath Howell, Head of Innovation, reporting savings of four minutes per call.
‘With 50,000 inbound calls a month, that’s huge,’ says Kath Howell, adding that there are also downstream benefits as the detail and consistency of the summaries means other teams can use them, whether that’s for training or handling complaints, as well as improving the customer experience.
‘We definitely chose the right use case to build our confidence because it was simple and low risk,’ says Kath Howell, pointing out that agents can edit the summaries and overwrite, or add detail to them, if required.
‘It was a prompt-based use case that did not need a lot of additional development, and it was light touch from an architecture point of view,’ adds Kath Howell. ‘And because we’re using open-source solutions wherever possible, it’s also cost effective and flexible so you feel you can experiment quite freely.’
Scoping out early use cases: the search for low-risk ROI
It’s a principle the company is continuing to deploy, seeking out other use cases that provide good value for money if they work and a learning opportunity even if they don’t. The approach is more akin to a technology start-up than a staid insurer – Kath Howell says the company likes to be thought of as disruptive and innovative – and enables rapid learning in order to build inhouse capabilities and quickly demonstrate value.
‘As part of one major transformation, we had activities we thought would take years down to months’
There’s plenty of scope to deploy in the back office too, not least accelerating the transformation of legacy systems by rewriting and updating old code and finding smarter ways of doing things.
‘We had one major transformation that we thought would take years and was down to months,’ says Sarah Self of Aviva. ‘That was a real ROI.’
Even so, when it comes to showcasing ROI, it’s front-end applications that still compel because they not only cut costs but also grow the topline through better customer experiences. The underwhelming performance of chatbots, for example, is ripe for an AI intervention to transform self-serve experiences.
‘Customers can on occasion find chatbots frustrating but if we could crack this, there’s a huge ROI and opportunity to improve the customer experience up for grabs’
‘Customers can on occasion find chatbots frustrating but if we could crack this, so that using the chatbot is easy, intuitive and can solve customer queries, then there’s a huge ROI and opportunity to improve the customer experience up for grabs,’ says Sarah Self of Aviva.
Responsible AI: finding the balance
Much has been written about the risks of AI – the hallucinations, the rogue chatbots and the risks of bias and model drift – but understanding and managing risk is bread and butter for insurance companies.
‘For insurers, AI governance is like the weather. It’s a day-to-day conversation’
‘AI governance is so important and for insurers, it’s like the weather,’ says Prathiba Krishna, AI and Ethics Lead at SAS UKI. ‘It’s a day-to-day conversation for them.’
She says responsible AI means asking not whether an AI can be done, but whether it should be done. This is when it’s so important to have had early conversations about AI ethics and data privacy. SAS, for example has six core principles underpinning its approach to trustworthy AI: human centricity, inclusivity, accountability, transparency, robustness and privacy & security. Responsible AI doesn’t mean no AI but rather finding the right balance and right controls so that fears about risk don’t constrain innovation to the point where the company gets left behind in the AI arms race – which could be the biggest risk of all.
For Kath Howell at Markerstudy Group, the solution has been to embed data privacy, InfoSec Compliance, and risk teams into the working group from the outset.
‘They’re enthusiasts and pro-innovation,’ she says. ‘These teams design their governance process as we build our proof of concept, so the governance framework is grounded in experience and has evolved as the complexity of our use cases evolved.’
As always, the over-arching message is to keep the human in the loop. ‘Throughout our journey, we have worked closely with Microsoft, who sum it up perfectly when saying: ‘The technology should be a co-pilot, not an autopilot,’ comments Kath Howell.
This means no blackbox surprises. AI-powered decision-making must be transparent and explainable, with codified guard rails and clear evidence trails.
‘Everybody needs to remember that if it was illegal, unethical and immoral to do something without AI, that still applies now.’
‘You need to be able to demonstrate how what you did delivered a better outcome for the customer,’ says Sarah Self at Aviva. ‘Everybody needs to remember that if it was illegal, unethical and immoral to do something without AI that still applies now.’
As well as automated bias-checking tools, companies need to educate their teams on bias and model drift.
‘Everybody understands that these models can be biased because they are trained on all the good and all the bad,’ says Self. ‘We need to have awareness across the organisation about this, making sure people apply a level of critical thinking, questioning does this look right, how am I testing this.’
This is particularly important when it comes to real-time fraud detection. Whilst AI is a potent tool in the battle against fraud, Prathiba Krishna at SAS says moving to real-time checks creates additional issues.
‘You have to make sure there’s ethical confidence when it’s real time,’ she says. ‘With batch processing, you have time for more monitoring and extra checks.’
And with regulators still developing their frameworks, companies need to make sure their AI ethics and controls are fit-for-purpose both now and in the future.
‘The regulators are paving the way for ethical AI strategy, so don’t wait for it to land – make sure you’re heading in the right direction’
‘Companies need to prepare sooner rather than later,’ says Prathiba Krishna at SAS UKI. ‘The regulators are paving the way for ethical AI strategy, so don’t wait for it to land, make sure you’re heading in the right direction and have time to prepare.’
Scaling: the big challenge
Targeted use cases in specific areas of the business can have a significant impact, but by no means reflect the true potential of this technology. Unlocking this potential will require companies to scale pilot projects into full production and to embark on an enterprise-wide AI-powered transformation.
‘Not all insurers in the UK are at scale-up,’ says Prathiba Krishna at SAS UKI, noting that it’s easy to focus on the technology and then forget what matters to make it work at scale. ‘Is your workforce ready to go? Have you got all your training materials? Have you been upskilling across the organisation?’
This is key, because scaling is dependent on employee willingness to use the new tools. Despite understandable fears about impacts on headcount, however, many people are naturally curious and keen to learn about this headline-making technology.
‘We were expecting a degree of anxiety, but it was met by a lot of enthusiasm,’ reports Kath Howell of Markerstudy Group. ‘There was also some healthy scepticism when we started testing with users, but that was helpful because you do need to think carefully about this and have those checks and balances.’
Mindsets are important, with the industry needing to get comfortable with new ways of working. Some use cases fly, while others have to be reworked and relaunched, and many more that may stay stuck in the sandbox. And sometimes there has to be push back: on occasion business units may want a Rolls Royce AI solution for something that could be fixed using cheaper non-AI tools. This is all part of the learning process, and requires careful management, excellent communication and ongoing education across the enterprise to be sure everyone understands how these projects work and how they can be rolled out in a safe and responsible manner.
Timelines can vary. ‘With one case, we went from idea to working it up with the business through to genuine scale in five months,’ says Sarah Self of Aviva. ‘Sometimes it’s longer, because you have the use case a bit wrong and have to go back and iterate that.’
While it’s important to move thoughtfully and with proper attention to the risks, Self cautions against ‘analysis paralysis’.
‘When it comes to scaling, sometimes you have to be brave’
‘When it comes to scaling, sometimes you have to be brave, accept that you will have to iterate and have confidence in your guard rails and controls,’ she says.
Unlike the companies who contributed to this article, so many others are still stuck with smaller use cases and proof of concept pilots, so it’s no wonder that 9 in 10 senior tech decision makers admit they do not fully understand GenAI’s potential impact on business processes.
‘There are capabilities available now... that nine months ago we would have said were five years away’
‘There are capabilities available now in the tools and models that nine months ago we would have said were five years away,’ says Sarah Self of Aviva.
Even technology experts are on a learning curve. ‘We all need some level of education on this,’ says Prathiba Krishna at SAS. ‘And while it’s okay not to know, you have to be willing to learn.’
The good news is that when insurers open up education and upskilling opportunities, people are hungry to learn, recognising that this technology is the future even if they come with their own questions and concerns.
‘There’s an organic natural appetite to see what we are doing and to get involved’
People sense that the world is on the cusp of massive AI-driven transformation, both at work and in their personal lives. Insurers have an opportunity to grasp this potential to reimagine the future of the industry, for their customers, their employees and investors, and for the benefit of society.
Footnotes
Kath Howell,
Head of Innovation
Markerstudy Group
Kath Howell is the Head of Innovation at Markerstudy Group. Her current focus is on building a GenAI capability for the group – designing the operating model, identifying and productionising use cases, designing the future tech stack, introducing ethical and responsible governance, as well as developing AI literacy and acceptance across the business and customer base.
Before joining Markerstudy, Kath ran her own consultancy business, has been Operations Director at a financial service charity, and worked for IT Consultancy leading IT transformation.
Sarah Self,
Generative AI Director
Aviva
A strong business and technology leader, who creates value through driving strategically important delivery and nurturing talent.
Sarah is a talented leader who specialises in Technology, Generative Artificial Intelligence, Transformational Change and Cyber Security. In a successful career Sarah has set strategic direction, delivered multi-year and international transformational change agendas (£100m+) and led large and diverse teams often in pressurised environments. Throughout her career Sarah has demonstrated a passion for people, drive for action and personal integrity.
Sarah has been with Aviva for 20 years and has built an in-depth knowledge of the business through a variety of roles – most recently as Generative AI Director and UK Chief Information Security Officer (CISO).
Sarah is an advocate for diversity and inclusion, actively sponsoring initiatives both within and outside of Aviva.
Prathiba Krishna,
AI & Ethics Lead
SAS UKI
Prathiba is an experienced Data Scientist with a rich background in the Insurance industry. With a Master’s degree in Operational Research with Applied Statistics and Risk, her passion takes form through seeing the varying applications of Machine Learning and AI techniques, and how they propel data scientists to build better models and solutions. Skilled in data analysis and modelling, she utilises SAS software and Open Source to assess and address problems within enterprise organisations. Prathiba is also an advocate of Ethical AI who promotes the value of having models built in a Responsible way.
About Insurance Innovators
Insurance Innovators is your one-stop shop for all the latest
insurance and financial services content from across the world. With a huge portfolio of conferences, webinars, executive roundtables and reports to choose from, Insurance Innovators provides our community with relevant and future-facing information.
To learn more visit insurance-innovators.com.
About SAS
SAS transforms how insurance works through the power of data and AI. Our industry specific solutions empower insurers to make smarter decisions in real-time, manage emerging risks, prevent fraud and create exceptional customer experiences boosting profitability. SAS offers a range of implementation options supported by an extensive catalogue of services to deliver quick wins and rapid return on investment. Lead with confidence in a rapidly evolving industry to stay ahead of change, innovate faster and respond to risks. SAS is used in 1,400 insurance companies worldwide, 90% of insurance companies in the Fortune Global 500 rely on SAS, 47 of the top 50 largest global insurers are relying on SAS.
To learn more, please visit https://www.sas.com/.
Stay current with Insurance Innovators
Subscribe to our newsletter to receive news, insights and special offers.