Abstract:
While AI in higher education is mostly discussed in terms of teaching and research, the operational side of science management – its rule-based, document-intensive, knowledge-dependent administrative work – remains overlooked. This article presents three domain-specific AI tools developed at the University of Kassel: KIRA (quality assurance of study programme documents), NOVA (doctoral procedures), and EKI (export control and research security). They share a three-pillar architecture: curated knowledge, prompt-based behavioural steering, and iterative feedback. These tools provide decentralised access to expert knowledge without decentralising responsibility for decision-making. The article reflects on efficiency, quality, empowerment, governance tensions, and a planned metabot ecosystem.
Über die Autorin:
- Nina Felgen,
Dr., Verwaltungsangestellte, Referentin für nationale Forschungsförderung, Stabsstelle Forschungs- und Graduiertenförderung, Universität Kassel - Diema Janakat,
M.A., Verwaltungsangestellte, stellvertretende Leitung der Promotionsgeschäftsstelle, Universität Kassel
- Heiko Wolf,
Magister Artium, Leiter Qualitätsentwicklung, Universität Kassel
