loading page

Harnessing Large Language Models for Research Institutions: an example based on NASA / JPL use-cases
  • +11
  • Steffen Mauceri,
  • Asitang Mishra,
  • Ryan M Mcgranaghan,
  • Ashish A Mahabal,
  • Lukas Mandrake,
  • Benjamin Smith,
  • Dustin J Graf,
  • Benjamin Nuerenberger,
  • Alice R Yepremyan,
  • Brian Wilson,
  • Amanda Towler,
  • Kay Y Pak,
  • Miles B Pellazar,
  • Daniel J Crichton
Steffen Mauceri

Corresponding Author:[email protected]

Author Profile
Asitang Mishra
Ryan M Mcgranaghan
Ashish A Mahabal
Lukas Mandrake
Benjamin Smith
Dustin J Graf
Benjamin Nuerenberger
Alice R Yepremyan
Brian Wilson
Amanda Towler
Kay Y Pak
Miles B Pellazar
Daniel J Crichton


In the era of information overload, how can research institutions eeectively manage and utilize the wealth of data at their disposal? State-of-the-art Large Language Models (LLMs) could hold the key. These models have the potential to revolutionize work processes at research institutions, aiding tasks such as ideation, summarizing text, generating code, and question answering. These applications could signiicant-ly enhance scientiic research and knowledge discovery in earth and space science, aid decision making, and enable better alignment of challenges with the right experts. We conducted a survey of current thrusts and use cases from diverse user groups across the Jet Propulsion Laboratory (JPL), using a comprehensive approach that combines quantitative and qualitative research methods. Focusing on both available capabilities from industry providers and custom-built solutions, we'll share strategic recommendations for harnessing LLM capabilities and discuss their implications for governance workkows. We also explored how LLMs can enhance knowledge management and discovery by making complex scientiic information more accessible and easier to analyze. By leveraging the unique capabilities of LLMs, JPL and other research institutions can accelerate scientiic discovery and technology development in earth and space science .
18 Dec 2023Submitted to ESS Open Archive
27 Dec 2023Published in ESS Open Archive