Collaborative Development of Ethical Standards for Generative AI Use in Warsaw's Public Administration

This proposal outlines an innovative project to develop ethical standards for Generative AI (GenAI) use by employees of the city of Warsaw. Our approach demonstrates a collaborative, bottom-up process for establishing technological guidelines in public administration through collective decision-making. The project brings together diverse stakeholders: Warsaw's city administration, a university, an NGO, and a firm specializing in public policy design. This multi-sector collaboration ensures a comprehensive approach to creating ethical AI guidelines tailored for the public sector. Our methodology features: 1. A participatory approach engaging office employees as experts 2. An iterative design process: analysis, workshops, and standard development 3. Education on GenAI, legal, and ethical aspects for the working group 4. Risk mapping of potential AI applications in the office 5. Solution prototyping of guidelines and principles 6. Expert consultations for solution verification 7. Resident involvement through opinion-gathering workshops 8. A holistic approach from needs identification to implementation planning 9. Transparency in communicating the project's process and results The process focuses on establishing standards that address the intersection of AI technology and urban governance, balancing ethical considerations with stakeholder engagement in AI adoption. The outcome is a framework document, emphasizing principles such as augmenting human workers, serving the common good, and responsible technology use. Crucially, this document is designed to be regularly updated and adapted to the dynamically changing world and technological developments. By showcasing this collaborative process and its adaptive outcome, we aim to contribute to the broader discussion on integrating AI into public administration, positioning Warsaw as a model for addressing AI-related challenges through inclusive, stakeholder-driven approaches.