UrbAI: Exploring the Possibilities of Generative AI Image Processing to Promote Citizen Participation

Research Background and Issues

• Issues and Challenges:

  • Current urban development offers limited opportunities for citizen participation, discouraging broader contributions of ideas and lacking intuitive methods for expressing concepts.
  • Existing approaches (e.g., written submissions, petitions, face-to-face workshops, or using augmented/virtual reality) fail to achieve high-fidelity idea representation, leaving the public reliant on professional tools used by experts (e.g., architects) to express their views.
  • High-fidelity technologies (e.g., CAD tools) are generally inaccessible to the public due to high costs and the need for specialized training.

• Significance:

  • Public input is critical in urban development to ensure transparency and sustainability in society.
  • Early-stage citizen participation is particularly essential for enhancing public acceptance and fostering innovation.

• Research Motivation:

  • The rise of Generative Artificial Intelligence (Generative AI, GenAI) offers convenient new pathways to enhance public engagement in urban development through visualization.
  • Related studies indicate that visual elements play a significant role in stimulating reflection, communicating complex issues, and facilitating collaborative design.

Solution

• Proposed System: UrbAI:

  • The system introduces a user-friendly generative AI tool allowing citizens to take photos of urban landscapes with their smartphones, annotate areas needing improvement, and provide descriptive suggestions, ultimately generating high-fidelity images (visualized modified ideas) automatically.
  • It offers multi-functional capabilities, including image processing, a community discussion platform, and feedback modules, enabling users to deeply engage and iterate on their proposals.

• Innovative Features:

  • Democratized Tools: Eliminates the need for citizens to possess design skills, lowering the technological entry barrier.
  • Collaborative Creation: Positions GenAI as a co-creation tool for citizens, encouraging progressive exploration from problem identification to solution development.
  • Feedback and Transparency: Provides feedback (e.g., compliance, feasibility, cost, environmental impact) to help citizens understand the context and constraints of their proposals.

• Key Steps and Technologies:

  1. Citizens upload images, and AI scans for potential urban issues (e.g., noise pollution, safety hazards).
  2. Users describe new improvement ideas through multimodal inputs (text, voice, or annotated photos).
  3. The system offers feedback based on municipal regulations and environmental adaptability, aiding further refinement of proposals.
  4. After submitting concepts, ideas are categorized and made visible on a community forum, allowing other residents and experts to evaluate, comment, and collaboratively improve them.

Research Outcomes

• Specific Findings:

  • Technology Probe Study (N=16):
    • Compared to traditional written formats, GenAI tools more effectively stimulated citizens' creativity and imagination.
    • During application, citizens shifted their focus from problem description to attempting solutions.
  • Expert Interviews (N=6):
    • GenAI was considered a powerful method for enhancing early-stage citizen participation, attracting more non-professionals.
    • Concerns were raised about excluding elderly individuals or those unfamiliar with digital technologies.
  • Prototype Design and Evaluation (N=10):
    • Citizens generally found UrbAI's step-by-step guidance and image-generation feedback to enhance process operability.
    • The system's community features (viewing, building opportunities for collective improvement) were highly inspiring, helping citizens understand each other's needs.

• Comparison with Existing Solutions:

  • Achieved seamless transitions from low-fidelity to high-fidelity representations, overcoming language/sketch expression barriers in traditional methods.
  • Provided more transparent feedback, strengthening interactions between citizens and local governments.

• Experimental Evaluation Results:

  • Generative AI significantly enhanced users' abilities in creativity, vision expression, and understanding feasibility.
  • Community-level functionalities facilitated more collaborative and inclusive discussions.

• Limitations and Future Directions:

  • Limitations:
    • Digital tools may exclude individuals unfamiliar with technology.
    • High-fidelity images may constrain participants' imagination or create unrealistic expectations.
    • The accuracy or diversity of content generated by GenAI tools may be limited by their training data.
  • Future Directions:
    • Explore integrating GenAI into face-to-face workshops to benefit technology novices.
    • Optimize generative models to better accommodate input patterns from ordinary citizens, evolving from demand-driven functionality to broader scenario capabilities.
    • Design hybrid methods (e.g., combining low-fidelity prototypes with high-fidelity images) to optimize exploratory and flexible dialogues in design processes.

Conclusion

This study demonstrates how Generative Artificial Intelligence (GenAI) can serve as a collaborative tool in urban development to engage citizens, enhance creativity, and promote public participation in decision-making. Specifically, the study identifies several key principles for designing citizen-friendly AI tools, such as transparency, ease of use, and creative support. Future research should focus on more effectively integrating generative AI with traditional urban planning methods to maximize benefits for all stakeholders.