67471 research outputs found
Sort by
Generative Agents at Work: Redesigning Administrative Processes at the German Federal Employment Agency
Get PDFThe integration of Generative Artificial Intelligence (GenAI) in public administration offers new ways to handle service and technology complexity while meeting high standards. This paper presents a case study from the German Federal Employment Agency, where a multi-agent system using local large language models (LLMs) automates the conversion of information technology (IT) change requests into structured IT development tasks (Jira tickets). Specialized agents interpret requirements, break them into tasks, and generate consistent entries. This improves organizational efficiency, reduces routine work, and outperforms traditional automation by enabling context-aware reasoning and dialogue, all within a secure, on-premise environment. While motivated by Germany’s demographic challenges, the findings have global relevance for public sector modernization and automation
Toward Smart City Digital Twins: Bridging Preparation and Response Phases in Urban Flood Evacuation through Agent-Based Modeling
Get PDFThis study presents an agent-based digital twin model for urban flood evacuation, using Okazaki City, Japan, as a case study. The model bridges the preparation and response phases of flood management by combining static infrastructure data with dynamic flood hazard and evacuation instruction data, enabling simulation of individual-level evacuation behaviors under changing conditions. By quantifying journey durations, the model revealed disparities in accessibility to safety. Results show that evacuation success depends on infrastructure and individual factors, with limited mobility increasing the risk of delay or failure. It also supports scenario-based evaluation of policy options, such as evacuation strategies and road closures, enhancing applicability across different disaster types and urban contexts. While the current version relies on secondary data and simplified assumptions, it demonstrates the feasibility of using digital twins for evacuation planning. Incorporating behavioral and traffic data and engaging stakeholders more deeply would enhance the model’s realism and strengthen its role as a reliable mirror of real-world evacuation processes for policy evaluation
Vision-Language Models (VLMs) in GeoAI Systems: Enhancing Brownfield Change Detection through Semantic Reasoning
Get PDFThis paper presents a hybrid GeoAI methodology for semantic change detection in satellite imagery by integrating Vision–Language Models (VLMs) into an unsupervised clustering-based pipeline. Building on earlier work using K–Means clustering to validate pre-selected brownfield regions from SPOT (2021–2023) and aerial imagery, we address the persistent issue of semantic ambiguity, particularly in high-uncertainty zones. We introduce a zero-shot, reasoning-based verification layer that evaluates whether visual differences across time are structurally meaningful. This approach improves interpretability, traceability, and diagnostic robustness. Evaluation across 1,000 human-labeled samples and a temporally unseen test set of size 200 (SPOT23–SPOT24) demonstrates notable improvement in ambiguous zones in reasoning quality and error transparency, especially where prior methods faltered. Our framework maintains the speed and scalability of clustering while injecting semantic precision through natural language decision paths. Designed with the UN’s SDG 11 (Sustainable Cities and Communities) in mind particularly for brownfield redevelopment this work contributes to scalable, interpretable, and operationally viable GeoAI systems
Bridging the Gap: A Systematic Review of Cyber Conflict Forecasting Models and the Case for AI-Driven Dynamic Frameworks
Get PDFCyber conflict forecasting remains constrained by static models that overlook the integration of geopolitical context with technical indicators. This systematic literature review examines 58 studies (2010–2025) using PRISMA guidelines and an Input-Process-Output framework to classify approaches and identify key gaps. Quantitative methods dominate (67%), yet only 14% incorporate geopolitical variables, despite the political nature of cyber conflict. Major limitations include adversarial adaptation blindness (85% assume static behavior), coarse temporal granularity (72% use daily+ intervals), lack of uncertainty quantification (75%), and minimal modeling of cross-domain escalation (92% cyber-only focus). Strategic forecasting is rare, with just 14% providing long-term insights and 16% offering decision support. In response, we propose eight design principles for AI-driven frameworks, emphasizing multimodal integration, adaptive threat modeling, fine-grained temporal analysis, and human-AI collaboration. This work lays the groundwork for dynamic forecasting systems that better support proactive cyber defense strategy and national security planning
Patterns in the Silence: Understanding the Differences in the Intent, Intensity, and Reach of State-Induced Internet Shutdowns
Get PDFDisrupting digital networks has quickly become a core strategy within autocratic governance, critical for effectively stifling dissent and preventing anti-regime mobilisation. However, the scope within which one achieves such disruption is broad. Treating such a strategy as a monolithic behaviour, therefore, would be a mistake, as there is substantial variation in how states utilise and harness this phenomenon, as well as to what ends. Yet, while all events are unique, it would be equally misguided to treat shutdowns as an endlessly fragmented series of isolated, idiosyncratic events outside of a broader understanding of repression. This study theoretically systematises and empirically demonstrates how internet shutdowns can be understood as existing within patterned but flexible dimensions of repressive intensity, distribution and precision. These dimensions recognise the shared modalities, operational logics, and functions of shutdowns across cases, allowing for the development of differentiation between them
Introduction to the Minitrack on Generative AI in IS Research and Education: Opportunities and Challenges
No full textIdentifying Human-GenAI Relations for Knowledge Management with Action Design Research Approach
Get PDFRecognizing the flexibility and autonomy of interactions between humans and Generative AI (GenAI), the conventional service system development approach faces challenges, and a new approach is needed to identify the roles of both users and GenAI. In this study, we adopted the Action Design Research（ADR） approach to develop a knowledge assistant (KA) that facilitates knowledge activities occurred in communities of practice. Lead users co-designed anticipated KA roles, aiming for enhancing role congruence between humans and KAs. Periodic interviews, tracing evolving human–KA relationships, provide insights to refine functions and strengthen collaboration. Using content analysis leveraged by ChatGPT’s natural language processing capabilities, we validated distinct patterns of role enactment between users and their KAs. This study demonstrates a new development cycle that emphasizes role congruence in developing human–AI collaborative service systems, particularly in relation to the socialization stage of the knowledge creation process
Do Investors Trust in AI Investments of European Companies?
Get PDFAnnouncements of emerging technologies often lead to notable stock market reactions, with Artificial Intelligence standing out due to its transformative potential and growing regulatory attention. Yet, most research on investor responses to AI disclosures focuses on U.S. firms, leaving the distinct European context unexplored. Using a short-term event study of 526 AI-related announcements by STOXX Europe 600 firms between 2015 and 2024, we report a significantly negative average stock return of -0.176% within a three-day window. However, announcements detailing specific AI technologies, involving collaborations with AI specialists, or made after the release of ChatGPT are associated with less negative reactions. In contrast, references to EU regulatory frameworks like the AI Act show no significant effect. Our findings confirm generally negative investor reactions to AI announcements but show that in Europe, strategic factors such as announcement specificity, collaborations, and timing also significantly mitigate these effects