Abstract
Designing artificial intelligence (AI) using participatory design (PD) methods is becoming fundamental as AI increasingly augments everyday life. Well documented cases of machine bias, where AI systems informing hiring, loan approvals and prison sentencing have discriminated against people with certain demographics (i.e. gender, race), have highlighted the need to engage end-users in AI design. PD methods show promise in designing AI systems for maximum societal benefits as they allow users to collaborate with researchers and make decisions about how AI systems should be designed. This paper presents a modified virtual world café method, based on the World Café method, as a PD method to identify relevant design requirements for designing AI systems from the beginning of development. Specifically, we describe how this method was developed and used to create design requirements for sound sensing AI for the home with UK-based residents. Findings suggest that the method can be used as a PD method at the beginning of AI system development to define design requirements. The paper concludes with reflections on how the virtual world café method performs as a PD method for designing AI and how findings might be taken forward in future PD research for designing AI systems more generally.