Branching Foresight - A Novel Interaction Concept for AI-generated Scenario Exploration

Paper Title

Branching Foresight - A Novel Interaction Concept for AI-generated Scenario Exploration

Publication Info

• Topic area: Climate coping and interactive AI scenario planning
• Keywords: Climate transilience, AI-generated scenarios, human-AI interaction, coping strategies, iterative branching, multimodal contextualization, reflection prompts, user experience, creativity support, climate communication

Background and Problem

• Problem / challenge: Climate futures often feel cognitively distant and emotionally overwhelming, making it difficult for individuals to translate concern into constructive action. Existing tools either require facilitation or fail to integrate emotional regulation and coping mechanisms effectively.
• Significance: Addressing climate anxiety and fostering coping capacity is critical for enabling individuals to persist, adapt, and transform in the face of climate challenges.
• Motivation and related work: Prior approaches in participatory foresight and AI-driven scenario generation lack accessibility for individuals, integration of reflection mechanisms, or empirical evaluation of psychological outcomes. This paper seeks to bridge these gaps by introducing a self-guided, emotionally supportive system.

Solution

• Proposed approach: Branching Foresight, a self-guided tool that combines AI-generated branching scenarios, multimodal contextualization, and embedded reflection prompts to cultivate climate transilience.
• Novelty:
  1. Integration of adaptability, persistence, and transformability mechanisms into a unified interaction loop.
  2. Use of iterative branching to rehearse alternative futures and scaffold coping strategies.
  3. Evaluation of psychological outcomes (transilience and climate-related emotions) in a single-session, self-guided format.
• Procedure and key techniques:
  • Users input personal context to seed AI-generated scenarios across three horizons (+5, +10, +15 years).
  • Scenarios are paired with images and conversational agents for guided exploration.
  • Reflection cards capture coping confidence and concrete actions, feeding into a results dashboard.
  • The system architecture includes a generation layer (GPT-4 for text, Stable Diffusion for images), a state layer (branching tree and reflections), and a presentation layer (map, timeline, chat interface, and results dashboard).

Results

• Concrete findings:
  • Significant pre-post gains in climate transilience: Adaptability (+1.7), Transformability (+1.3), Persistence (+1.0), Overall (+1.3).
  • Medium-to-large reductions in climate-related negative emotions: Climate Change Anxiety (-0.20) and Hopelessness (-0.39).
• Advantage over baselines: Demonstrates feasibility for short-term psychological improvements without requiring facilitation, unlike traditional scenario planning methods.
• Experiments / evaluation:
  • Study 1 (formative walkthrough): Identified usability barriers and refined design.
  • Study 2 (single-session pre-post): N=30 participants; measured transilience, climate emotions, user experience (UEQ-S), and creativity support (CSI).
  • Behavioral telemetry tracked dwell time, chat activity, and reflection confidence.
• Limitations and future work:
  • Single-session design without a control group; effects may reflect novelty or demand characteristics.
  • Uneven plausibility of generated futures; personalization challenges with free-text input.
  • Future directions include immersive VR environments, tangible interfaces, and expanded branching logics.

Summary

Branching Foresight integrates iterative branching, multimodal contextualization, and structured reflection to support climate transilience. In a single-session study, participants showed significant gains in perceived coping capacity and reductions in negative emotions, alongside positive ratings of usability and creativity support. Orientation tools (map, timeline, results dashboard) anchored clarity, while reflection prompts helped translate futures into actionable steps. The system bridges gaps between traditional participatory foresight and AI-generated narratives, offering a scalable, self-guided approach to climate coping. Future work aims to enhance realism, extend interaction modalities, and explore broader applications beyond climate change.