The Power of AI Language Models in Unstructured Data

Join Mark McCahill as he describes how advances in AI language models are making it feasible to search, summarize, and extract meaningful information from unstructured / semi-structured text and documents. New AI capabilities can augment how we handle and process information that does not fit well into tabular formats like spreadsheets or relational databases -- for example: grant proposals, textbooks, scientific publications, etc. This presentation will cover how Semantic Search, Knowledge Graphs, and Retrieval Augmented Generation (RAG) technology can extend the capabilities of Large Language Models (LLMs) such as ChatGPT by reducing LLM's tendency to "hallucinate" incorrect information. We will also discuss some potential applications of these AI technologies at Duke.