Open Access

Kommunale Kläranlagen stehen zunehmend vor der Herausforderung, Blähschlammereignisse unter sich verändernden klimatischen Randbedingungen und schwankenden Zulauffrachten sicher zu beherrschen, ohne auf energie- und chemikalienintensiven Betrieb ausweichen zu müssen. In dieser Studie wird ein hybrider Bilderkennungsalgorithmus vorgestellt, der klassische Bildverarbeitung mit Deep-Learning-Methoden (YOLOv4-basierte Objektdetektion) kombiniert, um in mikroskopischen Phasenkontrastaufnahmen von Belebtschlamm Flockenmasse, Filamentkonzentration und morphologische Merkmale von Mikroorganismen automatisiert zu quantifizieren. Der Algorithmus wurde in einer kommunalen Kläranlage mit sporadisch auftretenden Blähschlammereignissen unbekannter Ursache evaluiert. Hierzu wurden über mehrere Monate täglich Mikroskopiebilder erhoben, mit Betriebs- und Wetterdaten verknüpft und zwischen einem stabilen Referenzzeitraum und einem problembehafteten Zeitraum mit Blähschlammereignissen verglichen. Weder die vom Algorithmus ermittelten Filamentparameter noch die parallel durchgeführte 16S-rRNA- und ITS-Ampliconsequenzierung zeigten einen Anstieg filamentöser Mikroorganismen, eine Vergrößerung der Flocken oder signifikante Unterschiede in Alpha- und Beta-Diversität zwischen Referenz- und Problemzeitraum, sodass eine klassische filamentöse Überwucherung als Ursache ausgeschlossen werden konnte. Der Algorithmus erweist sich damit als robustes Werkzeug zur verlässlichen Detektion von Null-Ereignissen und zur ökologisch sinnvollen Entscheidungsunterstützung, indem er unnötige Fällmitteldosierungen vermeidbar macht und als Baustein zukünftiger Frühwarnsysteme auf Basis automatisierter Bildanalyse dienen kann.

Automating CVSS scoring can reduce analyst workload, but practical deployment requires explanations that are not only interpretable, but also faithful to model decisions. We present XplainCVSS, a framework that predicts all CVSS v3.1 base metrics from NVD vulnerability descriptions and evaluates explanation faithfulness quantitatively. Using 173,202 CVEs with strictly temporal splits, we benchmark ten models from TF-IDF+XGBoost to domainspecific and multi-task transformers. Fine-tuned SecureBERT achieves the best performance (macro-F1 = 0.774; reconstructed base-score MAE = 0.934), while multi-task variants offer competitive consistency at 7–10x lower training cost. We validate TreeSHAP and gradient-based transformer attributions across all eight CVSS components, showing that highlighted features are decision-relevant. A temporal drift decomposition further shows that label-distribution shift (concept drift) is the main driver of degradation (delta-F1 = 0.079), indicating the need for periodic retraining. Riskaware evaluation yields a low 2.79% under-prediction rate for Critical vulnerabilities, supporting analyst-in-the-loop triage. Overall, XplainCVSS highlights a practical trade-off between transformer accuracy and the direct interpretability of keyword-based models.

Artificial intelligence (AI) is rapidly reshaping public administration by influencing how governments design services, manage data, and exercise regulatory authority. National AI strategies have become essential tools for governments to articulate their ambitions and define their role in governing AI, yet the policy landscape has evolved significantly since their adoption. This paper examines how national AI strategies define the role of public administration in AI governance. Drawing from concepts in AI governance, digital government, and policy design, we propose a role-based analytical framework that conceptualizes public administration as AI adopter, regulator, and orchestrator of AI ecosystems. We apply the framework in a qualitative comparative policy analysis of national AI strategies in Germany, the Netherlands, and the United Kingdom. Using a deductive-inductive qualitative content analysis, we reconstruct country-specific role profiles and compare strategic emphases across the three cases. Our findings reveal that all three strategies foreground the adopter role, with extensive public-sector use cases and capacity-building measures, while systematic scaling beyond pilots is made explicit only in the UK’s “scan, pilot, scale” approach. Regulatory framings are more prominent and tightly coupled to EU-level developments in Germany and the Netherlands, whereas the UK emphasizes innovation and competitiveness with fewer detailed legal commitments. Orchestration arrangements likewise diverge, ranging from partnership-oriented ecosystem coordination in the Netherlands and the UK to more hierarchical steering in Germany. We argue that a role-based lens helps to uncover how early strategic choices position public administrations within emerging AI ecosystems and to identify misalignments between these framings and the evolving European AI governance landscape, thereby providing an empirically grounded lens to assess whether existing strategies remain fit for purpose.

Fremdkörper auf Start- und Landebahnen (Foreign Object Debris, FOD) stellen ein erhebliches Sicherheits- und Kostenrisiko für den Luftverkehr dar. Trotz bestehender Inspektionspflichten erfolgt die Erkennung in der Praxis überwiegend manuell, zeitaufwendig und personalintensiv. Eine flächendeckend eingesetzte, zuverlässige automatisierte Lösung existiert bislang nicht. In einem Forschungsprojekt an der Technischen Hochschule Wildau wurde daher ein autonomes, bodengebundenes Inspektionssystem entwickelt, das mithilfe kamerabasierter Objekterkennung FOD auf Flugplatzflächen detektiert. Das System ist auf kurze Inspektionszeiten, kleine Objektgrößen und den Einsatz unter realen Betriebsbedingungen ausgelegt. Erste Tests zeigen eine Detektionsrate von 97,7% bei 53 False Negatives bei einem Testdatensatz von 2.000 Bildern.

In the last few years Generative Artificial Intelligence (GenAI) has taken universities by storm. This does not only affect the way what and how to teach and how to assess knowledge and competences in exams, but it also impacts the way how to do research. On one hand, GenAI has fast access to a huge amount of wide-ranging (explicit) knowledge and the capability to extract and combine knowledge that seems to be relevant in the context of certain research. On the other hand, research is closely linked to the human capability of being creative showing curiosity, care, collaboration, and critical thinking. Against this background, the paper addresses a question of immense social and also ethical relevance: May we use (Gen)AI as tool for scientific research and if so to what extent? The paper starts analysing the problem from a theoretical and conceptual point of view by comparing capabilities of GenAI with those of a good researcher. Based on this the complete research cycle is investigated to understand which capabilities are more helpful to achieve research results of high quality and novelty, but in an effective way. To gather empirical data from a self-experiment, the setup of an unexperienced researcher (e.g. PhD student) with just a rough idea about the intentional research focus was chosen. Addressing the first and fundamental phase of any research process, idea generation, by comparing performance of an AI researcher with a human one, results demonstrate the usefulness of combining strength of both worlds mandatorily keeping the human researcher in the loop. With this, the paper focusses on chances and challenges coming with the highly dynamic evolution of AI within a complex, knowledge-intense application area where human experience, problem-solving capability and creativity is decisive. It contributes to discussions about necessary adjustments in how organizations (and particularly universities) manage their knowledge generating potential considering artificial and human intelligence.