StayFocused: Examining the Effects of Reflective Prompts and Chatbot Support on Compulsive Smartphone Use

Title of the Paper

StayFocused: Examining the Effects of Reflective Prompts and Chatbot Support on Compulsive Smartphone Use

Paper Information

• Subject Area: Human-Computer Interaction (HCI) and Behavior Change
• Keywords: Smartphone addiction, reflection, persuasive technology, conversational user interface, large language models, productivity tools, digital health, chatbots, behavior design

Research Background and Problem

• Problems and Challenges:

  • Smartphone addiction has negative impacts on physical health, mental health, and productivity, particularly among young people and university students who are more susceptible to over-reliance on technology.
  • Existing interventions (e.g., screen time monitoring, app restrictions) have limitations and fail to fundamentally address compulsive smartphone use.
  • There is a lack of in-depth understanding of users' immediate behavioral reflection in specific contexts (e.g., "non-screen time") and its long-term effects on smartphone use.

• Significance:

  • Gaining a deeper understanding of and reducing smartphone addiction, especially among young people, can help improve individual focus and task completion ability, thereby enhancing overall quality of life and mental health.

• Research Motivation and Related Work:

  • Previous studies have shown that reflective interventions can induce positive immediate behavior changes in areas such as healthy eating and exercise, but their effects on smartphone addiction have not been fully explored.
  • Chatbots have potential applications in health and behavioral interventions, but their use as tools for reflective interventions is still an emerging field, with unclear effects.

Solution

• Method:

  • Developed a mobile application called StayFocused, designed to help users reduce smartphone use during focus periods through reflective prompts and chatbot functionality.
  • The app includes three versions for comparative experiments:
    1. Baseline Version (PB): Supports users in setting focus periods but does not include reflective prompts or chatbot support.
    2. Reflection Version (PR): Displays a series of reflective questions when users exit focus mode or complete focus tasks.
    3. Reflection + Chatbot Version (PRC): Guides users through reflective questions and provides instant feedback via a chatbot powered by GPT-3.

• Innovations:

  • Combines "Situated Reflection" and "Reflection-on-Action" to enhance users' immediate motivation and long-term behavior change.
  • Explores the potential of large language models (e.g., GPT-3) in supporting user behavior reflection and emotional support.

• Implementation Steps:

  • Users set focus periods (ranging from 25 to 125 minutes) to initiate focus mode.
  • If users check their phones or attempt to exit during the focus period, the app issues reflective prompts or initiates a conversation via the chatbot.
  • After completing focus tasks, users answer questions about their focus experience.

Research Findings

• Specific Results:

  • A 5-week test involving 36 university students found that reflective prompts helped users extend focus time and enhance self-control.
  • Chatbot support further improved the effects of reflection, particularly in strengthening users' commitment to goals and providing emotional support.

• Comparative Advantages Over Existing Solutions:

  • Compared to solutions relying solely on time management or productivity tools, StayFocused better stimulates users' reflective awareness and sustains behavior change.
  • Beyond supporting behavioral reflection, the chatbot was perceived by users as an emotional and personable companion, increasing their willingness for long-term use.

• Experimental and Evaluation Results:

  • Users in the PR and PRC groups significantly increased their daily focus time compared to the baseline group.
  • PRC group users further reduced smartphone usage during the intervention phase, with the most significant reduction (approximately 14.36%).
  • The chatbot increased the length of users' responses, and sentiment analysis showed predominantly positive feedback.

• Limitations and Future Directions:

  • Limitations:
    • The chatbot exhibited some design flaws, including insufficient contextual understanding and limited question diversity.
    • The experimental group consisted of a limited population sample (primarily U.S. university students), raising potential cultural applicability concerns.
    • The small sample size weakened the interpretive power of behavioral changes between groups.
  • Future Directions:
    • Optimize chatbot prompt design to make questions more diverse and dynamically adjust dialogue content based on user progress.
    • Further enhance the chatbot's ability to handle complex interaction scenarios through fine-tuning of large language models.
    • Expand application scenarios to other behavior intervention areas, such as healthy eating, exercise tracking, and mental health support.

If the information provided is extensive or requires deeper interpretation, please let me know for further elaboration.