TU Graz
  • AI for Teaching

Prompt Transparency and Accountability in Organisational genAI Systems to Support Course Descriptions Improvement

Presentation at the STS Conference Graz 2026: how the TU Graz course description chatbot is built — and why disclosed prompts create accountability.

When a university runs an AI system in regular operation, the prompt helps decide what that system treats as good teaching. The presentation introduces the TU Graz course description chatbot and argues that disclosing those prompts is a workable route to accountability in organisational AI systems.

The contribution was prepared for the session “AI in public and private sector organizations: accountability, trustworthiness, transparency” at the STS Conference Graz 2026. It presents a genAI chatbot that supports TU Graz teaching staff in improving their course descriptions — the same system that is in use as the Course Description Chatbot.

The second part moves beyond the single case: in an organisational AI system it is not the model that makes the value judgements but the prompt guiding it — which quality criteria apply, how strictly they are checked, what counts as a shortcoming. If that prompt stays invisible, the assessment cannot be retraced and no one can be held answerable for it. The presentation therefore treats prompt transparency not as a documentation duty but as the mechanism that makes accountability attributable in the first place.

The presentation

How to cite

Brünner, B., Furch, M., Hoffmann, K., & Leitner, P. (2026). Prompt Transparency and Accountability in Organisational genAI Systems to Support Course Descriptions Improvement [Conference presentation]. STS Conference Graz 2026, Graz University of Technology. https://doi.org/10.3217/w2k5d-s1w36

Show BibTeX entry 
@misc{brunner2026prompttransparency,
  author       = {Brünner, B. and Furch, M. and Hoffmann, K. and Leitner, P.},
  title        = {{Prompt Transparency and Accountability in Organisational genAI Systems to Support Course Descriptions Improvement}},
  howpublished = {STS Conference Graz 2026},
  publisher    = {Graz University of Technology},
  year         = {2026},
  doi          = {10.3217/w2k5d-s1w36}
}

Related services

Participating partners

TU Graz

Within TU Graz

Graz University of Technology is a research partner in several projects on AI-supported learning and learning analytics.

← Back to publications

The logos of the partners and the projects are protected by copyright and trademark law and are excluded from this site’s CC BY licence. They appear here solely to identify the cooperation; all rights remain with the respective organisations.