Transforming Public Engagement through Generative AI

Tags: #Urban #Singapore #LargeLanguageModel #GenerativeAI 

Background

The Urban Redevelopment Authority sought to innovate visitor engagement through cutting-edge generative AI technology. The project aims to create an interactive, technology-driven experience that allows exhibition visitors to explore and visualize urban spaces through AI-generated imagery. 

Problem Statement

How can URA leverage generative AI to create an engaging, interactive exhibition experience that allows visitors to explore and personalize visualizations of urban sites, enhancing public understanding and interaction with urban planning?

Solution Approach

Key Technologies

• AI Image Generation: Flux.1 frontier AI model

• Hardware: iPad Air 13-inch (M2 chip)

• Interactive Interface Design

Innovative Elements

• Three-stage user interaction process

• AI-powered image generation with user-guided customization

• Inpainting technique for advanced image modification

Key Features and Benefits

Technical Capabilities

• Guided Image Generation

  • Users can select sites and keywords to generate custom images

• Multi-stage Interaction

  • Site selection

  • Keyword-based customization

  • Image preference selection

  • Rapid Image Generation

    • Four images generated within 10 seconds

Business Value

• Enhanced visitor engagement

• Interactive learning experience

• Innovative demonstration of AI capabilities in urban planning

• Collect user insights and preferences

Use Cases

Urban Visualization

• Scenario

  • Exhibition visitors explore potential urban designs

• User Type

  • General public, urban planning enthusiasts

• Functionality

  • Generate and customize images of urban sites based on user preferences

Interactive Learning

• Scenario

  • Visitors understand urban planning possibilities

• User Type

  • Students, professionals, community members

• Functionality

  • Create personalized visualizations through AI-guided interface

User Personas

Curious Visitor

• Name: Emily

• Background: Urban design student

• Challenges: Understanding complex urban planning concepts

• Solution Interaction: Uses AI interface to generate and explore site visualizations

Community Planner

• Name: Michael

• Background: Local community representative

• Challenges: Visualizing potential urban developments

• Solution Interaction: Experiments with different keywords to see site variations

Data and Integration

Data Sources

• URA site information

• User-generated inputs

• Flux.1 AI model

Integration Approach

• Comprehensive data collection in CSV format

• Tagged image generation

Technical Challenges and Solutions

• Challenge: Rapid, high-quality image generation

  • Solution: Utilizing Flux.1 AI model to ensure quick, relevant image outputs

• Challenge: User-friendly interface

  • Solution: Three-stage interaction with clear, guided steps

Results and Impact

• Performance Metrics

  • 1,000 projected user sessions

  • Four AI-generated images per session

  • Image generation within 10 seconds

• User Engagement

  • Interactive exhibition experience

• Innovation Impact

  • Demonstration of AI in public sector engagement

Conclusion

The URA's Generative AI Proof-of-Concept represents a innovative approach to public engagement, leveraging cutting-edge AI technology to make urban planning more interactive, accessible, and engaging for the general public.