Project 2: Identifying AI Risks for Non-Human Life in Urban Spaces

Back to all projects

AI systems used in streets, parks, and other urban spaces can change how non-human life moves, feeds, reproduces, and survives in cities. Yet existing work does not provide a clear, systematic map of AI-related harms across the full range of non-human life in urban environments.

We contribute the following:

• Compile a structured list of concrete AI uses in urban spaces.
• Develop a categorization of non-human life in urban spaces.
• Design a method that uses a large language model (LLM) to elicit risks by asking it to reason from the perspective of a selected non-human being (for example, a cat) for a given AI use.
• Release a dataset and analysis code so others can reproduce our results.

References

1. [1] Why AI Harms Can’t Be Fixed One Identity at a Time: What 5,300 Incident Reports Reveal About Intersectionality.