Cost Management for the Use of Generative AI in Higher Education: A Case Study from North Rhine-Westphalia

9 pages•Published: November 6, 2025

Abstract

Generative Artificial Intelligence (AI) models, like systems such as ChatGPT, have catalyzed a transformative shift in higher education, prompting universities to explore innovative applications across research, teaching, and administration. This paper discusses the KI:connect.nrw project, which aims to provide universities in North Rhine-Westphalia (NRW), Germany, with equitable access to commercial and open-source AI services while addressing the challenges associated with cost management. The project introduces a centralized web interface that facilitates user-friendly access to multiple AI systems through API integration. Key objectives include transparent cost allocation tailored to specific user groups or projects and flexible budget control mechanisms that accommodate diverse funding sources within participating HEI. The paper further examines the complexities of billing methods, ranging from account-based to token-based models, and their implications for the central provider and using universities. Initial experiences at RWTH Aachen University highlight the project's operational framework and its potential impact on managing AI-related costs effectively as usage among employees and students increases.

Keyphrases: ai, billing, budget control, cost management, llms, web interface

In: Laurence Desnos, Raimund Vogl, Lazaros Merakos, Carmen Diaz, Janina Mincer-Daszkiewicz and Stuart Mclellan (editors). Proceedings of EUNIS 2025 annual congress in Belfast, vol 107, pages 108-116.

Links:	https://easychair.org/publications/paper/lTvF
	https://doi.org/10.29007/43wh

BibTeX entry

@inproceedings{EUNIS2025:Cost_Management_Use_Generative,
  author    = {Bernd Decker and Denise Dittrich},
  title     = {Cost Management for the Use of Generative AI in Higher Education: A Case Study from North Rhine-Westphalia},
  booktitle = {Proceedings of EUNIS 2025 annual congress in Belfast},
  editor    = {Laurence Desnos and Raimund Vogl and Lazaros Merakos and Carmen Diaz and Janina Mincer-Daszkiewicz and Stuart Mclellan},
  series    = {EPiC Series in Computing},
  volume    = {107},
  publisher = {EasyChair},
  bibsource = {EasyChair, https://easychair.org},
  issn      = {2398-7340},
  url       = {/publications/paper/lTvF},
  doi       = {10.29007/43wh},
  pages     = {108-116},
  year      = {2025}}

Download PDF Open PDF in browser