AI Language Models in the Global Insurance Industry

How will artificial intelligence (AI), particularly large language models, impact the insurance industry in 2024?

Conversations around artificial intelligence (AI) have permeated all aspects of our society in 2023. Insurers are increasingly trying to get more specifics on how AI will impact the broader insurance industry and their lines of business. Large language models (LLMs) have been a source of particular interest and confusion. Insurance technology executives express a spectrum of views on these tools, ranging from questioning whether an insurer should build LLMs to questioning if LLM capabilities have any value whatsoever.

Insurers Have The Choice of Several Paths When It Comes to LLMs

Insurers have a range of approaches to choose from when they engage with generative AI:

  • Insurers who want to develop and control their LLM may choose to build their own. This path will likely be rare due to the prohibitive cost of building one.
  • Other insurers will choose to do some fine-tuning on a preexisting LLM, which involves refining the model with partners to create an offshoot model focused on a specific area.
  • Insurers may also leverage language model capabilities from a proven insurance solution provider or a new insuretech player in the space.
  • Lastly, insurers may choose not to pursue the use of LLMs at all, perhaps building a small language model instead and using open-source tools for specific limited purposes.

Datos Insights expects fewer than 1% of insurers to build LLMs and around 5% of insurers to fine-tune an LLM. Most insurers will leverage a solution provider’s AI capabilities or partner with insuretech companies.

Many Insurers Are Not Engaging With This New Technology

Despite widespread interest in AI and related technologies, many insurers are not engaging with newer technology, instead sticking to uses that pre-date 2023, such as expert systems, machine learning, and machine visioning. However, insurers who are not investing any effort or mindshare into these topics do risk placing themselves at a competitive disadvantage. Few insurers are likely to build their own data science foundations, but understanding the concepts and options around AI will be necessary for insurers to leverage, invest in, or effectively work with external entities.

The Decision of Whether or Not to Leverage LLMs Should Be Strategic

No matter which path insurers choose when it comes to LLMs, the decision should be strategic, with specific use cases outlined where the use of the technology will add value. There are several different strategic parameters they will need to consider, including budget, insurer data science expertise, insurance domain knowledge, and competitive advantage.

For more information about the value and future of AI language models in the global insurance industry, read our full report on the subject, The Value and Future of AI Language Models in the Global Insurance Industry, October 2023, sponsored by DXC.