Accelerating Scientific Discoveries: NASA’s Use of Large Language Models for Unlocking the Value of Science Data and Information

Artificial Intelligence (AI) is transforming the scientific research process and has the potential to accelerate scientific discovery and analysis. Generative AI, specifically large language models (LLMs), allows scientists to ask questions, solve problems, and generate novel ideas, overturning long-held scientific workflows and ideals. Tools such as Elicit and NotebookLM demonstrate potential in revolutionizing these scientific workflows while emerging frameworks such as the AI Scientist push the paradigm even further by enabling fully automatic scientific discovery steps. Recognizing the potential of these emerging techniques, NASA is investigating the development and adoption of LLMs to help scientists, industries and governments unlock the value of NASA’s open data and information. In this presentation, we explore the development of a large language model tailored to NASA's needs, leveraging curated data from diverse scientific fields. We also share lessons learned from the development of the model, applications that are using the model and our vision for the future of scientific discovery in a world enabled by LLMs.