From Text to Self: Users’ Perception of AIMC Tools on Interpersonal Communication and Self

Document Title

From Text to Self: Users’ Perception of AIMC Tools on Interpersonal Communication and Self

Document Information

• Subject Area: Human-Computer Interaction (HCI), Artificial Intelligence (AI), and Computer-Mediated Communication
• Keywords: AI-Mediated Communication (AIMC), Interpersonal Communication, Diary Study, Digital Communication, Language Models, User Attitudes, Interaction Design, Social Perception

Research Background and Issues

Identified Problems or Challenges

• With the rapid proliferation of AI-Mediated Communication (AIMC) tools powered by large language models (LLMs), this technology is significantly transforming the way interpersonal communication occurs.
• Existing studies suggest that AIMC tools may have both positive impacts (e.g., improving communication efficiency and positivity) and negative effects (e.g., reduced trust and diminished social perception).
• To date, there is a lack of research exploring how users employ these tools to express themselves and their perceptions of the user experience.

Importance of the Research

• Interpersonal communication not only affects individual quality of life but also has profound implications for professional relationships, team collaboration, and the accumulation of social capital.
• The widespread adoption of AIMC tools may have long-term effects on societal communication patterns and individual cognition, making it theoretically and practically significant to explore their pros and cons.

Research Motivation and Related Work

• Based on a review of literature and practical issues regarding how AI influences human communication, the authors propose refined research questions:
  • RQ1: What support can AIMC tools provide for interpersonal communication?
  • RQ2: How do users evaluate their experiences with these tools?
  • RQ3: In what contexts do users accept these tools?
  • RQ4: How do users perceive the long-term impact of these tools on their communication practices?
• The article draws on extensive related literature, including theories of computer-mediated communication (e.g., hyperpersonal communication theory) and the impact of AI tools on trust and self-presentation.

Solutions

Methods and Research Design

• Mixed Methods: The authors conducted a one-week diary study and semi-structured interviews, recruiting 15 participants in total.
  • Diary Study: Participants recorded at least three instances of interpersonal communication modified using AIMC tools daily and completed a questionnaire.
  • Interviews: Pre- and post-study interviews were conducted to understand users’ attitudes toward tool support and gather design suggestions.
• Research Process:
  • Recorded users’ original messages, AIMC-generated outputs, final sent versions, and reflections.
  • Collected data on user satisfaction, emotional experiences, tool usage frequency, and outcome acceptance.

Innovations

• Proposed a "Communication Space" framework that categorizes communication scenarios along two dimensions: communication risk (high or low) and relationship dynamics (formal or informal).
• Analyzed the suitability of the tools for different communication scenarios and the variations in user satisfaction.

Research Findings

Key Findings

1. Strengths and Weaknesses
   • Strengths: Boosted users’ communication confidence, enhanced precision in expression, supported cross-cultural communication, and facilitated emotional understanding.
   • Weaknesses: Outputs were often verbose, rigid in language, overly emotional, and could lead to feelings of inauthenticity and guilt among users.
2. Satisfaction and Context Suitability
   • Satisfaction showed significant variation: tools were more suitable and received higher satisfaction in formal, high-risk scenarios, while they were less effective or deemed unnecessary in informal, low-risk scenarios.
   • Tools were perceived as "optimizers" in formal contexts and as "quick assistants" or "tone checkers" in informal settings.

Experimental Evaluation and Results

• The average user satisfaction score was 7.1 out of 10, with satisfaction increasing gradually after a short learning period.
• Qualitative data revealed users’ multifaceted attitudes toward communication tools, while quantitative analysis highlighted contextual differences in satisfaction.

Limitations and Future Directions

• Limitations:
  • Participants had to access the tool via a web interface, which may have reduced user experience.
  • Data relied heavily on self-reported measures, which could introduce selection bias.
  • The study duration of one week was insufficient to observe long-term effects.
  • The sample size was limited, with only 14 participants completing the study.
  • Some results may be biased due to limitations of mainstream tools (e.g., ChatGPT).
• Suggestions for Future Research:
  • Explore behavioral changes resulting from long-term use.
  • Consider personalized customization and contextual support (e.g., relationship dynamics and communication goals).
  • Investigate whether AIMC tools can genuinely enhance offline communication skills.

Design Recommendations

1. Global Customization and Real-Time Support

   • Provide customizable user profiles to dynamically adjust tool outputs based on context.
   • Enable real-time language analysis and instant feedback functionality.

2. Enhancing Personal Style Retention and Optimization

   • Allow users to upload personal text data to improve consistency between tool outputs and users’ original style.

3. Controlling Information Length and Emotional Intensity

   • Offer settings to adjust the length and tone intensity of output text.

4. Multidimensional Evaluation Mechanism

   • Provide tone and sentiment analysis to predict recipient reactions and support users in setting and tracking communication goals.

5. Transparency and Educational Features

   • Improve the tool’s explanation mechanisms to help users better understand its functions and logic, supporting language skill development.

Conclusion

• AIMC tools can effectively enhance communication quality in specific scenarios, but issues such as verbose outputs and excessive emotionality may reduce user satisfaction.
• Designers should focus on balancing users’ needs for linguistic, emotional, and interactional preferences, particularly between personalized expression and practicality.
• Future research and system development should continue to examine the long-term effects of these tools on interpersonal communication to ensure ethical and socially responsible technology design.