The ERGO Innovation Lab has published a roughly 30-page white paper on AI applications and their underlying language models. With the “ChatGPT and large language models: An introduction with a view to the insurance industry” guideline, several ERGO experts from the areas of machine learning, conversational AI and innovations share their extensive knowledge of large language models (LLMs). In individual chapters, they explain in detail both the functionalities and potentials of the models, as well as their possible risks and limits. The authors also use concrete practical examples to show how texts can be produced, dialogues conducted or data classified with corresponding tools. This is true of both the insurance industry and other industries.
Jens Sievert, Innovation Manager at ERGO Innovation Lab, says in this regard: “AI applications such as ChatGPT and the like have huge potential for the insurance industry, where every day there are hundreds of thousands of customer touchpoints in which language has to be processed. Insurance companies are therefore predestined for the use of language models and language-based tools. But the insurance industry is also one of the most heavily regulated industries that there is, which is why the use of technologies has to be carefully considered and implemented. With our white paper, we therefore give an initial, comprehensive orientation and show what companies must pay attention to here.”
ERGO has great expertise in the use of language models and rule-based voice assistants. For example, the ERGO Advanced Analytics and AI Unit already uses LLMs to classify and extract information from documents. The ERGO Voice Unit has in turn been designing rule-based phonebots for Customer Service on the basis of artificial intelligence since 2020, and is already working on a GPT-based prototype. In the ERGO Innovation Lab in Berlin, GPT language models and ChatGPT applications are also being tested for their potential in a dedicated “opportunity space”.
All this practical knowledge has gone into the white paper that has now been published and is available for all interested parties to download free of charge.