We designed and evaluated an AI-based Application to enhance human creativity in design thinking workshops. The results indicated that AI hindered human creativity, resulting in fewer idea generations. The findings from quantitative and qualitative analyses comparing the only-human and human-AI teams indicated that AI contributed to the usability of ideas during the divergent phase and supported humans in converging on more novel ideas. The further development of the application is necessary to consider how humans can collaborate with AI without relying on it.

This work introduces a graph-based CAD assistant that predicts the next modelling operation in parametric design sequences. Real CATIA V5 models from the automotive domain are converted into directed acyclic graphs capturing feature dependencies, enabling learning directly from structural design data. A four-layer Graph Attention Network achieved a top-5 prediction accuracy of 94%, outperforming a frequency-based non-parametric baseline. The results show that graph representations and attention-based message passing provide a strong foundation for context-aware modelling assistance.

This study examines how engineers formulate natural language prompts for searching existing assemblies in mechanical design. A survey with 48 engineers produced 169 prompts for different assemblies. Results show that prompts are mostly written as bullet points with an average of three and up to seven requirements. The engineers describe assemblies mainly through implicit functional descriptions and geometric or physical parameters. These findings form an empirical basis for developing generative AI-driven, prompt-based systems to foster design reuse.

Current approaches for the generative design of sheet metal parts only take singular optimization goals into account. This paper presents a concept for a deep reinforcement learning approach to train an agent to generate sheet metal parts by combining segments from a predefined library. Through a weighted reward function, agents can be trained for different or combined optimization goals, such as weight, cost, or sustainability. The resulting agents enable the creation of a pareto front of optimal solutions, supporting efficient exploration of the design space for diverse design objectives.

Generative Artificial Intelligence (GenAI) is transforming design practice yet research lacks empirical insights into adoption in real-world design organisations. Through the case study of a European automotive OEM, we found that GenAI could accelerate ideation, but adoption was limited due to critical concerns regarding intellectual property, data security, originality, and the risk of skill atrophy. Thus, organisational capabilities like workflow specific training, transparent governance of data protection policies, and cohesive toolchains are needed for successful GenAI integration.

Agile teams often encounter obstacles in systematically identifying the underlying root causes of collaboration challenges and deriving effective countermeasures. Grounded in the Design Research Methodology, this study investigates a hybrid AI-human approach for targeted generation of problem-specific reference and impact models to enhance systematic improvement in agile product development. A structured workflow integrates AI capabilities (e.g. scaffolding, consistency) and expert knowledge (causality, context), while a three-stage review ensures methodological rigor and result reliability.

A technology-oriented approach to AI predominates in research and practice, yet despite a high level of technological readiness, projects often fail due to poor domain-specific problem framing and data quality in early-stage AI system development. This contribution conducts an analysis of existing AI-related readiness models, to identify gaps in addressing these factors. The use case-centered AI readiness level framework is proposed on the basis of these findings – a unified, evidence-based model that links problem, data, and technology readiness across planning and implementation stages.

Collaboration is crucial in design and management, fostering innovation, problem-solving, and decision-making. We explore the use of vision-language models (VLMs) for analyzing collaboration, focusing on detecting social behavior and group affect. By fusing multimodal cues, VLMs enable more context-aware reasoning beyond surface-level perception. We develop a pipeline, a structured prompt and an interactive visualization for integrating VLMs into the analysis workflow. Comparing VLM and human analysis results, we discuss how VLMs can advance collaboration analysis and the remaining challenges.

As digital mental health interventions expand, integrating EU regulations into the design process remains essential but challenging due to their complexity. This study explores how the GDPR, AI Act, EHDS, HTA, MDR, and IVDR influence the design of AI-based mental health chatbots by mapping them onto a framework. The proposed mapping approach provides an overview of the regulatory landscape at each stage, revealing tensions between innovation and compliance as well as opportunities to use regulatory principles as structured checkpoints that guide responsible digital mental health design.

Requirements quality shapes engineering design, yet natural language specifications remain vulnerable to ambiguity. We investigate how LLMs support ambiguity detection using a hybrid dataset combining NASA JWST requirements with systematically injected defects. Using auto-extracted domain knowledge, we compare a domain-agnostic baseline with a context-aware approach. Incorporating domain knowledge helps LLMs better distinguish genuinely ambiguous requirements from acceptable ones, highlighting the potential of context-aware AI assistants for requirements engineering and early-stage design.

Why do all design acts begin by explicating a bounded frame of work? In design ontology, framing is the selection and representation of components and features in a system to guide perception and decision of designers but remains implicit. As a structural abstraction it becomes an explicit principle, formalised by a computational methodology that parameterises bounds and projects elements of a design having weighted attributes, in a relational context. Thus, the cognitive act becomes epistemic to compute for generating and evaluating frames, aligning design reasoning with scientific discourse.

Learning MBSE is hindered by abstraction and complex tools. This paper identifies barriers via literature review and interviews to design a RAG-based chatbot acting as a “smart view” for contextual guidance. Evaluated through a semester-long field study and a controlled experiment, the prototype shows high usability and reduces cognitive load. While performance is comparable to traditional e-books, the RAG-enabled system effectively mitigates entry-level barriers and aids authentic project work through stepwise tutoring, offering a scalable, interactive complement to MBSE education.

Early integration of circularity into complex system architectures remains ad hoc and weakly verifiable. Model-based systems engineering (MBSE) can address this issue with precise modeling and early verification. This paper presents an artificial intelligence supported MBSE method to integrate circular strategies based on SysML v2 models, highlighting potential changes and enhancements. Generative suggestions are integrated via a regulated workflow with traceable justifications and context-enhanced queries. A use case modeling the print head of a 3D printer illustrates this approach.

This study investigates how large language models (LLMs) support extracting technical requirements from early product pitches. Mechanical engineering students worked under three conditions: manual, LLM-assisted, and LLM combined with a QFD interface. Both AI-assisted conditions improved requirement quality and lowered perceived difficulty. Thematic analysis showed cognitive effort shifted from generating requirements to evaluating and verifying AI outputs, while the LLM-only group reported the most positive attitudes.

Trade-off studies often use the design of experiments approach, while simulation models enable data-based product optimization by AI. This paper presents a comparison of evolutionary algorithms, reinforcement learning as well as active learning for design space exploration. Based on a real-world case study and hypervolume analysis, the performance of selected algorithms is assessed. The results highlight their ability to identify pareto fronts and provide insights to deepen the understanding of AI-driven design space exploration.

This paper introduces Dignity-Centered Design (DCD), a sociotechnical framework for AI-mediated systems. While AI ethics often focuses on concepts such as fairness and transparency, DCD evaluates how systems shape lived experience, power dynamics, and human agency. Drawing on healthcare traditions and the Dignity Index, the framework articulates three dimensions (individual, relational, and systemic) alongside five core principles. It includes a Dignity Spectrum in AI System Design to assess design choices and applies these to healthcare AI to support reflective practice.