Today, Generative AI and LLMs (Large Language Models) represent a major area of innovation in the insurance and actuarial fields. These models offer promising perspectives for transforming operations, services, and risk understanding.
Projects may focus on simple operational tasks. In that case, LLMs are used for data quality improvement, information extraction, text classification, question-answering systems, summary, etc. But these projects may also be related to more advanced topics related to risk assessment like underwriting risk evaluation, claims knowledge graph, etc.
In any case, and despite their potential, putting these technologies into production remains a challenge: model complexity, lack of adaptation to specific domains, unmet business expectations, IT constraints, costs and maintainability that are difficult to master, etc.
For these reasons (and many others), it seems increasingly necessary to have a solid culture in LLM applied to Insurance. The aim of this workshop is to:
- Acquire a solid understanding of the origins, development, and technical properties of Large Language Models, with a particular focus on their application in the insurance and actuarial sectors.
- Learn how to effectively transform and represent data to optimise its use in LLMs, understanding the importance of cleansing, segmenting, and embedding.
- Develop practical skills for using LLMs, including prompting, fine-tuning, and handling various tasks such as data extraction, classification, and summary creation.
- Understand methodologies for evaluating, developing, and deploying LLM projects, with a clear understanding of operational challenges, security, and quality.
- Explore and apply the knowledge gained through real-life case studies, focusing on solving problems specific to the insurance sector, to reinforce learning and encourage innovation.
