Large Language Models - Autograding & Governance for businesses

back to overview

Type and Duration

FFF-Förderprojekt, September 2023 until July 2025

Coordinator

Data Science & Artificial Intelligence

Main Research

Business Process Management

Description

This project aims to investigate the potential of large language models (LLMs), specifically ChatGPT, in two areas: exam grading support and governance of LLMs in an industrial context.
The first aspect focuses on utilizing LLMs for grading exams, addressing concerns such as subjectivity and time-consuming manual grading. The project aims to develop a methodology for educators to verify grades using LLMs while considering ethical considerations, biases, and hallucination issues of LLMs. It evaluates LLM responses for short textual answers in multiple languages and discusses the challenges of fully automating grading.
The second aspect examines the governance of LLMs in a business context. While AI regulation has received attention recently, the governance of LLMs within organizations has been understudied. The project analyzes data from local businesses and relevant literature to understand governance challenges related to LLM hallucinations, intellectual property risks, and potential benefits. It aims to develop a framework for effective LLM governance.

Publications

  • Schneider, J., Meske, C., & Kuss, P. (2024). Foundation Models. Business & Information Systems Engineering. (ABDC_2022: A; ABS_2021: 2; VHB_3: B)

    more