Document Type

Masters Project

Abstract

This study examines how generative artificial intelligence (AI) can enhance efficiency, accuracy, and compliance in university-based research administration amid increasing federal oversight and workforce constraints. Baseline findings indicate that administrative burden is driven less by time and more by complexity, multi-source dependency, and workflow inefficiencies, particularly rework. Using a structured, mixed-methods approach, this research evaluates the application of large language models (LLMs)—ChatGPT, Google Gemini, and Microsoft Copilot—across core workflows, including funding opportunity analysis, award terms extraction, closeout requirements, and budget development. A controlled, iterative testing framework was implemented using standardized prompt structures, with outputs evaluated against human-prepared benchmarks for accuracy, completeness, consistency, usability, and regulatory alignment. Findings demonstrate that AI improves workflow efficiency when implemented within a structured framework incorporating prompt standardization, iterative refinement, and human-in-the-loop validation. Performance differences across platforms were secondary to prompt design. Results indicate that AI is most effective as a decision-support tool that reduces rework and supports interpretation of complex regulatory requirements rather than as a fully autonomous solution.

Publication Date

5-1-2026

Download

Share

COinS