Mark Crovella Awarded Project from First-of-its-kind NSF Backed Program to Study Large Language Models

By Andrew Thurston, The Brink

Faculty of Computer & Data Sciences Professor Mark Crovella‘s new project studying potential biases of AI was one of 35 projects announced as a National Artificial Intelligence Research Resource (NAIRR) Pilot Project. This new NSF-backed program grants Crovella access to advanced supercomputing resources and data in order to investigate if AI models incorporate biases against certain groups or tend to spread extreme views. The NAIRR projects came to be from President Biden’s October 2023 Executive Order aiming for a coordinated approach to “governing the development of AI safely and responsibly.”

Mark Crovella is a professor in the Department of Computer Science and a founding member of the Faculty of Computer & Data Sciences. Crovella will utilize the access granted by NAIRR to study large language models (LLMs) — AI algorithms equipped to summarize, translate, predict, and generate text and data — and determine whether these models contain biases, promote hateful speech, or spread disinformation, with the goal of making these models more trustworthy. LLMs are currently a part of ChatGPT as well as generative AI chatbots like Microsoft Copilot, and are rapidly spreading to other sectors like education.

Learn more about the project in The Brink’s interview with Mark Crovella here.