The Sound of Support: Gendered Voice Agent as Support to Minority Teammates in Gender-Imbalanced Team

Title of the Paper

The Sound of Support: Gendered Voice Agent as Support to Minority Teammates in Gender-Imbalanced Team

Paper Information

• Research Domain: Interpersonal Interaction and Human-Agent Team Collaboration
• Keywords: Human-Agent Team Collaboration, Voice Agent, Minority Group, Majority Group, Gendered Design, Marginalization, Artificial Intelligence, Interaction Design, Social Presence, Gender-Imbalanced Teams

Research Background and Problem

• Problem or Challenge:
  • Minority group members in gender-imbalanced teams are more likely to experience isolation, difficulty in building rapport, and limited supportive feedback.
  • Gender disparities and job satisfaction issues for women in STEM fields persist.
  • These challenges negatively impact minority group members' performance and team experience.
• Significance:
  • Gender-imbalanced teams are prevalent in educational and workplace settings and require urgent solutions. Addressing the marginalization of minority group members can foster team harmony and improve overall performance.
• Research Motivation and Related Work:
  • Existing literature suggests that machine agents (e.g., robots or intelligent assistants) can positively influence teams through social presence.
  • Research on human-agent team collaboration indicates that agents can contribute to balancing team interactions and creating a more inclusive team environment.
  • However, there is a lack of focus on direct support for minority groups, and the potential of technology to support minority groups remains underexplored.

Solution

• Proposed Method or Solution:
  • Employ gendered voice cues in intelligent agents participating in team collaboration to provide support to minority group members.
• Innovations:
  • First exploration of the potential of same-gender voice agents in supporting minority groups.
  • Extends traditional interpersonal group behavior theories to human-agent team collaboration.
• Implementation Steps and Key Techniques:
  • Experimental Design:
    • 2 x 2 x 2 experimental design: Participants were assigned to gender-imbalanced teams based on their gender and randomly paired with either a female or male voice agent to complete tasks.
    • Sample size: 178 participants, with teams categorized as female-dominated and male-dominated.
    • Task content: Brainstorming tasks and team decision-making tasks.
  • Data Collection and Analysis:
    • Self-report questionnaires were used to measure team experience, perceptions of marginalization, and views on human and AI teammates.
    • Participants' conversational behaviors were recorded to analyze contribution proportions and the number of ideas generated.
    • Controlled variables: Task order, team composition, agent gender, etc.

Research Findings

• Specific Findings:
  • Same-gender agents significantly supported female minority group members: They contributed more ideas and increased their speaking proportion in the team.
  • Male minority group members were more talkative when collaborating with male agents but showed reduced focus.
  • Behavioral outcomes of same-gender agents and minority group members were similar to those of human teammates.
• Advantages Compared to Existing Solutions:
  • Provides a low-interference, non-awkward technical support solution.
  • Avoids overly anthropomorphic designs, presenting agent gender identity solely through voice cues.
• Experimental or Evaluation Results:
  • Female minority group members experienced reduced feelings of marginalization and improved performance.
  • Positive changes for male minority group members were limited, with no strong preference for agent gender.
• Limitations and Future Directions:
  • The solution offers limited support for individuals who have experienced severe marginalization.
  • Gendered design uses binary classification; future studies should explore more inclusive designs.
  • Small experimental scale and short-term tasks require further research on long-term team collaboration.
  • Suggest expanding research to more natural workplace/educational scenarios to observe real-world team dynamics.

This study highlights the potential of voice agents as a support mechanism for minority group members, but addressing gender imbalance fundamentally requires long-term practice and institutional changes.