Funding Source: NSF #2326174 – In the past two years, artificial intelligence (AI) tools, such as ChatGPT, have revolutionized the software industry. These models could be particularly valuable in the natural and social scientists by enabling scientists who are not expert programmers to turn descriptions into computer programs. However, current models do not support the programming languages or practices commonly used by scientists.

This project aims to make AI programming tools accessible for scientists. To understand scientists’ needs, Anderson and Teich will run qualitative and quantitative studies of how scientists write programs. These findings will help them develop benchmarks and models for text-to-code generation that focus specifically on the programming paradigms used by scientists. This grant will also fund an investigation of how students who learn to program in science courses interact with AI programming tools. By harnessing AI to make programming easier for scientists, this project aims to accelerate scientific discovery and allow more people to participate in scientific work.

Faculty: Carolyn Jane Anderson and Erin Teich
Department: Computer Science
Funding Source: NSF #2326174