Reinventing Insurance with GenAI

It’s barely 18 months since Open AI’s large language model (LLM) – ChatGPT– exploded into public consciousness. Yet within this timeframe, the conversation among many insurance companies has progressed rapidly from how they might apply GenAI in discrete pilots to how they can implement it at scale to achieve the huge potential value on offer.

According to Microsoft research, 41% of business leaders say that within the next five years they’re going to fundamentally redesign their business processes around AI. The 27th PwC CEO survey also highlights the imperative to act, with one in five business leaders saying that their business will not be sustainable within 10 years if it remains on its current path. Across the insurance industry, the focus on experimentation is now shifting to industrialisation. That reflects the growing urgency to harness GenAI’s potential.

However, that’s not to say there aren’t some big challenges to overcome along the way. Precisely how best to move from the pilot and proof-of-concept stages to implementation of GenAI-enabled tools at scale is far from resolved. But some insurers are leading the way, and starting to see the benefits.

At the PwC Microsoft GenAI breakfast event, leaders from technology and insurance sat down to discuss how far GenAI has already progressed in the industry and, crucially, where its development is likely to lead in future.

No single, end-to-end use case – yet

With potential applications right across insurance processes, from underwriting to claims analysis and customer service, insurers are identifying the elements of processes where GenAI can create efficiencies, greater speed and accuracy. But rather than working on a singular large use case, insurers are using AI in a more incremental way. Capabilities such as summarisation, fact extraction, assisted content generation, and information retrieval are being inserted into processes to help people do their jobs, with the emphasis on augmentation rather than replacement.

But to achieve that cost effectively requires a clear structure and method. Creating each use case individually, carrying out the risk assessment, testing and implementing all drive up the unit cost per solution. To address those potentially unacceptably high costs, it’s important to develop a new approach that builds on a smaller number of foundational patterns that encapsulate a significant proportion of most models. With those in place – in the form of an ‘AI Factory’ – it becomes much easier to capture value at scale across the organisation.

With a huge and growing range of third-party GenAI tools coming on the market all the time, deciding which to use and when to build in-house is an increasingly critical decision. The rate of change makes this even more pressing.

In addition, as a regulated industry, insurers have to make sure that any potential use of AI is subject to strict risk assessment. Doing so on a case-by-case basis will add significantly to cost. So, it’s important that governance and risk management are built in rather than models being left for risk assessment once they’ve been created. And as the global regulatory landscape continues to evolve, insurers need to keep a watching brief on the implications for the transparency and explainability of the models they deploy.

Quality and quantity of data are, of course, foundational to GenAI models delivering the right results. It’s crucial to understand the data that is required for a specific application, and to assess its availability. Progressing to implementation at scale raises the data stakes even higher. But it’s also important to make sure that data does not prevent the vital experimentation and testing of use cases. It’s a tricky balancing act that every insurer needs to constantly navigate.

The human stays in the loop

GenAI is a revolutionary technology. That’s clear. But technology is only one element of insurers’ drive to harness its potential. People are fundamental to successful implementation. And that plays out in a number of ways. First is a clear recognition of the new skills that GenAI-enabled insurers will need to develop in their workforce. Take contact-centre enquiries, for example. GenAI chatbots can answer many different types. But they will struggle to address more complex and nuanced questions. That’s where people come in. However, with routine and repetitive tasks covered by GenAI, contact-centre staff will have to be able to answer customers’ more challenging questions. They’ll need upskilling to do that.

People are also essential to developing the uses of GenAI beyond the technology. Most of the innovation and ideas that will disrupt processes - for example underwriting - will come from people who run those processes today. Their input will be essential. But in order to make the most positive contributions that they can, they need to understand the technology’s capabilities. Exposing users to co-pilots and assistants embedded in everyday tools can help here. Thinking about the user experience is crucial too, so that GenAI tools become part of day-to-day operations. But more fundamentally, as insurers think about building their technology infrastructure for GenAI they also need to build a talent infrastructure that will help to capture GenAI’s potential today and into the future.

Wider horizons: change from operations to industry

As insurers move ahead with bringing GenAI use cases to life, they also need to keep one eye on the horizon for the larger changes that are likely to come in the future. At the moment, GenAI’s impact is largely at the operational stage. It’s transforming the way that processes, or parts of processes, are executed. But much more profound developments are on the way. GenAI will also enable new business models that are not feasible without AI. There are any number of possibilities. It might include expense ratios significantly lower than today’s; it could be new hybrid advice models emerging in the life and pensions markets; or it might mean algorithmic underwriting in the Specialty market.

Eventually disruption will likely be seen at the level of the market itself, changing the whole paradigm of how a discrete market operates. For example, in the not-too-distant future, a GenAI agent could take a customer’s details and, on their behalf, go and find the best motor policy at the best price from hundreds of different providers. Insurers have to start preparing for the possibility of GenAI ushering in disruptive change orders of magnitude greater (and faster) than anything the industry’s seen to date.

Of course, there are also emerging new risks that come with the broader use of GenAI, and insurers must also start thinking about the innovative types of products they could write to cover these.

Overall, GenAI is creating change in the insurance industry that’s at least as, if not more, far-reaching than any technology that’s come before it. To capture its potential, insurers need to understand how to scale it responsibly so that everyone – their people, clients, regulators and more – can all benefit from the value it can achieve.