Sage Grande: An Open Artificial Intelligence Testbed for Edge Computing and Intelligent Sensing

Language-based generative artificial intelligence (AI) is transforming scientific research, opening new opportunities to empower discovery and broader participation through the use of natural language interfaces to scientific workflows and AI-enabled cyberinfrastructure. To fully exploit NSF’s investments in cyberinfrastructure, there is a critical need to develop and understand the integration, programming, and use of AI across scientific computing resources and high-resolution environmental sensors and instruments such as cameras, microphones, and weather sensors.

This new AI-enabled cyberinfrastructure will accelerate data analysis and software generation with powerful multi-modal large language models (LLMs). It harnesses LLMs to democratize access to AI, enabling individuals without programming skills to conduct experiments and allowing the scientific community to develop natural language interfaces, empowering a new generation of scientists, broadening participation in AI research through outreach programs, and enhancing representation.

Hands-on training and educational resources for students will provide technical skills and enhance the nation’s AI workforce development.

The Sage Grande Testbed (SGT) is designed to revolutionize the integration of LLMs and edge computing in scientific research, extending NSF’s cyberinfrastructure to support advanced AI applications across various domains. It builds on SAGE, an AI-enabled platform developed through NSF’s Mid-Scale Research Infrastructure program to integrate natural language processing capabilities with cloud-based software environments, enabling seamless scientific inquiries and reducing the complexity of using advanced cyberinfrastructure.

SGT supports a library of LLMs to support natural language queries, fostering breakthrough research and hands-on education. The testbed addresses generative AI, the computing continuum, and scientific measurement and observation, and provides cyberinfrastructure to evaluate AI trust factors, including hallucinations, bias, and safety as well as AI-generated and AI-controlled cyber-physical systems necessary for field measurements and real-time data analysis.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.