Chatbots Facilitating Consensus-building in Asynchronous Co-Design

Document Title

Chatbots Facilitating Consensus-Building in Asynchronous Co-Design

Document Information

• Domain: Human-Computer Interaction, Co-Design, and Asynchronous Collaboration
• Keywords: Co-Design, Chatbots, Consensus Building, Asynchronous Interaction, Human-Computer Interaction, Decision Support, Collective Collaboration, Data Collection, Educational Research, User Research

Research Background and Problem

• Problem: Consensus building is critical to the success of co-design projects. However, managing group discussions in large-scale co-design projects with multiple stakeholders is challenging. Stakeholders must navigate conflicting needs and perspectives to align their opinions toward shared goals. Lack of experienced facilitators or resource constraints can lead to failures in consensus building.
• Significance: Without consensus, co-design projects may fail to achieve their objectives or even collapse. Furthermore, whether group decisions are accepted by stakeholders directly impacts project execution and subsequent progress.
• Motivation and Related Work: While existing research has explored how technology can support consensus building in long-term design projects (e.g., online discussion platforms and automatically generated design options), there is still a lack of focus on asynchronous co-design, particularly on how conversational agents or chatbots can mediate and guide such processes.

Solution

• Method Overview:

  • The authors propose a chatbot-based approach to facilitate consensus building among stakeholders asynchronously. The chatbot interacts with individual stakeholders to collect their needs, identify conflicts, and attempt to resolve them through dialogue.
  • The chatbot acts as an asynchronous facilitator, guiding stakeholders to collaborate without direct communication while protecting anonymity to encourage more genuine ideas.

• Innovations:

  • The chatbot is designed as a rule-driven system, adhering to established consensus-building guidelines and incorporating a dialogue-based conflict resolution process.
  • By leveraging asynchronous participation, the approach eliminates the challenges of coordinating real-time discussions faced by traditional facilitators.
  • The system displays other members' perspectives to enhance users' awareness of group interests and reduce discomfort from criticism or social pressures.

• Implementation Steps and Key Technologies:

  1. The chatbot introduces the purpose of the interaction and defines the conflict.
  2. It guides users to propose initial solutions and engage in self-reflection.
  3. Users are prompted to reconsider their stance from others' perspectives, with suggestions from other members displayed.
  4. Finally, users select the most satisfactory solution.
  • The process relies on rule-driven scripts and fixed dialogue flows rather than advanced natural language processing (NLP).

Research Outcomes

• Specific Results:

  • The chatbot significantly increased users' willingness to accept group decisions, even when these decisions conflicted with their personal interests.
  • The chatbot enhanced users' perception of collective effort and fair participation within the group.
  • Experimental results showed that among 12 participants, the majority believed the chatbot effectively facilitated conflict resolution and consensus building.

• Comparative Advantages Over Existing Solutions:

  • Compared to traditional methods, the chatbot enables asynchronous and anonymous interactions, expanding the scale of design projects while reducing social pressure during face-to-face discussions.
  • The system is simple and operates on existing chat platforms like Telegram, offering strong usability and scalability.

• Experimental and Evaluation Results:

  • Users did not engage in direct communication but still perceived the presence of other participants.
  • Participants reported that the chatbot improved their sense of group collaboration and fairness.
  • Despite final group decisions conflicting with personal opinions, most users were willing to accept and support them.

• Limitations and Future Directions:

  • The experimental scenarios were somewhat limited, as the conflicts were predefined by researchers, and the outcomes had minimal real-world social and economic impact.
  • Future work could integrate chatbots into multiple stages of co-design to explore their applicability in complex real-world projects.
  • Further research should enhance the chatbot's ability to personalize interactions with users, potentially incorporating more advanced NLP technologies.
  • The study also suggests developing more effective opinion aggregation and content filtering mechanisms to accommodate larger-scale group participation.