Conference Agenda

Life cycle assessments identify environmental hotspots, yet translating these insights into design actions remains slow and expert-dependent. Existing AI tools lack a dynamic link to current research. Here, we present an LLM-driven pipeline that interprets LCA hotspots, mines recent literature, and extracts feasible, research-backed design alternatives. In a case study on a headlamp control unit, the method produced relevant and applicable improvements, indicating its value for accelerating sustainable product design.

This study examines how AI can support the development of Leading Sustainability Criteria in sustainable product development, comparing AI-generated outputs with human-facilitated workshop results from four Swedish companies. Results highlight AI’s ability to accelerate and broaden sustainability framing, but emphasize that contextual relevance and legitimacy depend on participatory inputs. The findings suggest that AI is most effective when integrated into hybrid workflows that preserve human insight and stakeholder engagement—offering practical guidance for future implementation.

Barriers such as limited repair literacy and design-for-disposability continue to reinforce replacement cultures. This paper introduces AIFixer, an AI-powered interactive tool that guides consumers through electronic repair, promoting sustainable product lifecycles. Using a mixed-methods, user-centred approach, the study evaluates AIFixer’s usability and behavioural impact across real-world repair tasks. Findings show that conversational AI lowers barriers, builds confidence, and generates data for circular design, highlighting opportunities for multimodal and community-integrated development.

In a circular economy, repurposing extends product lifecycles and reduces resource use. However, identifying feasible repurposing opportunities remains challenging. This study therefore evaluates the capability of large language models to identify such repurposing scenarios and their relevant properties, using documented repurposing cases from peer-reviewed literature. Three models were tested, revealing potential in identifying repurposing scenarios, but also highlighting the need for structured methods and further research due to systematic limitations in property identification.