Cleaning & greening the city: AI-imaginations of positive safety

The growing role of artificial intelligence (AI) in public safety remains predominantly focused so-perceived ‘threats’ and ‘risks’. Far less attention is devoted to whether AI can be used to strengthen ‘care’ and ‘belonging’ from a Positive Safety perspective. This paper seeks to narrow this gap by empirically researching Lombardijen—a neighborhood in Rotterdam, the Netherlands, with notably low subjective safety perceptions. This study develops a Photovoice-to-AI methodology involving 60 high school students and 16 residents to document ‘safe/pleasant’ and ‘unsafe/unpleasant’ neighborhood spots, facilitating community AI-imaginations of their ‘ideal neighborhood’ in two workshops (±15 participants each). Guided by these results and community requests, we then iteratively engage 24 decisionmakers and nature representatives to gather their perspectives. The findings reveal two empirical AI-experiences of Positive Safety: (i) Care-oriented AI and (ii) Inclusive AI—especially relevant for ‘participatory cities’ like Rotterdam which can exacerbate inequalities. AI-imaginations overcome traditional communicative participation barriers, bringing overlooked experiences to light and making frictions between neighborhood stakeholders discernible; ultimately revealing that trash weakens—and nature strengthens—public safety perceptions in this neighborhood. We call for reframing trash as an invitation for care and nature, countering punitive Broken Windows policing approaches and underlining community-driven public space improvements that give agency to communities. We thereby aim to contribute to the debate on AI and public safety by rethinking AI design from a Positive Safety perspective based on virtues like care and trust, which bears promise for the Right to the City debate.