A recent survey by Celent reveals that by the end of 2023, half of insurers say that they’ll have tested generative AI (gen AI) solutions in the form of large language models (LLMs).
For a technology that has only been in the public eye for 12 months or so, the drive for adoption by the insurance industry has been remarkably rapid. But what we’ve seen so far is merely the tip of the iceberg, with gen AI’s impact only just beginning to make itself felt.
Looking forward, the potential for gen AI to transform the insurance industry is huge. It will be able to shoulder much of the burden of routine work – and more – that’s common in the industry today. Take claims assessment. An appropriately trained LLM will be able to interpret an event in the context of even the most complex contracts and determine a claim’s validity (or not) within seconds. This will allow, for example, many insurers to complete nearly 100% of missing claims information from complex submissions, which is expected to generate significant savings from greater operational efficiency and lower claims costs.
Transforming insurance front to back
Gen AI’s impacts will extend to many other areas of the industry. For example, it can be used to analyse vast quantities of data to provide simple, accurate summaries to underwriters as they make their assessments. Meanwhile, other functions, such as marketing, will also see gen AI completely change the art of the possible. It will be able to take standardized product and service content and blend it with personalized customer information to create truly bespoke communications at scale.
Because gen AI uses natural language for prompts and instructions, it democratizes access to insights that were previously only available to data scientists and specialists. It will also be put to work within technology departments too, writing code and scripts, and helping to support integrations.
In these contexts – and many others – gen AI will do the heavy lifting, enabling people to focus on business-critical tasks and activities that require the best of human innovation, empathy and creativity.
A clear case for change
It’s evident that the insurance companies that adopt gen AI the fastest will secure a considerable competitive advantage. The gains they could make are likely to fall into three broad categories:
1. Higher profitability and growth, by identifying currently untapped opportunities and enhancing products and customer experiences,
2. Cost savings from operational efficiency,
3. Operational intelligence and effectiveness from integrating gen AI into existing processes.
Navigating the challenges
Having said that, it’s also the case that many in the industry face challenges when it comes to moving from the current experimental phase to implementing gen AI at scale.
Why? By their nature, LLMs require significant quantities of well-managed, effectively organized, accurate and compliant data. And as a regulated industry, the compliance demands on insurers’ data exceed those of many other sectors, meaning insurers will need to ensure they continue to meet strict regulatory requirements for data privacy and stewardship. Integrating gen AI with existing legacy tech is another potential challenge. In fact, some 75% of executives across all industries cite this as an obstacle to progress. Here again, establishing a solid data foundation is a critical first step.
Other potential pitfalls associated with gen AI include so-called ‘hallucinations’ where gen AI effectively fabricates an answer. Biased outcomes that arise from gen AI learning from biases already inherent in the training data are a further common concern. Both, of course, need to be addressed. And they’re by no means uncontrollable. The best approach? Treat the outputs of an LLM with the same rigorous rules, policies and norms that any organization would apply to content created by a person. Building in the right controls around an LLM from the outset will help avoid many of the potential pitfalls.
Start now or fall behind
Insurers that have yet to start exploring the possibilities of gen AI need to get going soon. With their competitors already pushing solutions into production, it’s time to start identifying use cases and get to work building and deploying pilots to understand where the greatest benefits and value are likely to be found.
We list the best Large Language Models (LLMs) for coding.
This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
