As values shape the design and governance of technology, it becomes critical to move beyond universal framings to explore the nuanced, subjective understandings individuals hold about values. Traditional value elicitation methods often identify values at play but overlook how they are interpreted through individuals’ social identities and lived experiences. This paper introduces an AI-augmented value exploration method inspired by photo elicitation, which involves interviews supported by participant-taken photographs. Instead, we use AI-generated imagery to uncover hidden associations and insights around personal understandings of values. In an exploratory study with six participants, we focused on the value of well-being, examining how AI-generated visuals prompted diverse personal interpretations and facilitated deeper value reflections. Our findings show that this method uncovers implicit meanings and deepens discussions by translating abstract ideas into tangible interpretations to yield richer data on situated values.

As design research matures, more designers pursue a PhD. In its turn, the PhD itself is changing from a solitary preparation for a career in academia toward an increasing emphasis on interdisciplinary and international experience and a greater variety of jobs. These developments call for greater opportunities for PhD candidates to build their network during the PhD. The WunderLibrary, developed in the EU-funded project DoCS4Design, aims to connect PhD candidates through a web-based platform for sharing educational and research materials and making contact. To seed the platform's ontology, the collection of tags which connect the items in it, we studied the keywords and metadata from a set of 342 PhD theses from the 6 PhD programmes in the project over the last decade. The collection of theses came with between three and six freely chosen keywords, typically provided by the authors. In this paper, we describe the process of curating the keywords and clustering the resulting data on three levels. This produced a set of 342 keywords and two levels of clustering. The raw data are openly available. We discuss data analysis and a spin-off application that uses higher-level labels to help PhD candidates describe their work.