OK Google, Let’s Learn: Using Voice User Interfaces for Informal Self-Regulated Learning of Health Topics among Younger and Older Adults

Title of the Paper

OK Google, Let’s Learn: Using Voice User Interfaces for Informal Self-Regulated Learning of Health Topics among Younger and Older Adults

Paper Information

• Research Domain: Application of Voice User Interfaces (VUI) in informal, self-regulated learning of health topics
• Keywords: Self-regulated learning, aging, metacognition, informal learning, health information, interaction fluency, conversational agents, voice user interfaces, voice assistants, technology acceptance

Research Background and Problem

• Problems and Challenges: The potential of Voice User Interfaces (VUI) in learning health information remains underexplored, particularly for older adults. Overcoming voice interaction barriers in VUI (e.g., conversational interruptions) is a critical issue. Additionally, the suitability of learning methods across different age groups warrants investigation.
• Research Importance: With the widespread adoption of technology, VUIs (e.g., Google Assistant and Amazon Alexa) are increasingly integrated into daily life. Using VUIs for health information learning is not only convenient but also scalable, especially for older adults, as it holds the potential to support healthy aging.
• Research Motivation: Older adults are often underestimated and underrepresented in studies on technology use and learning participation. This study aims to explore the feasibility of self-regulated learning (SRL) through VUIs to meet the needs of older adults in accessing health information and learning.

Solution

• Method or Solution:
  • Developed a virtual voice agent called "Health Buddy" (Google Home Action) to test learning outcomes on health-related topics.
  • Designed learning modes based on three self-regulated learning (SRL) strategies:
    1. Monologue Mode (M): Primarily narrative-based, with minimal interaction.
    2. Conceptual Diagramming Mode (CD): Uses diagrams to help associate concepts.
    3. Scaffolding Building Mode (SB): Guides learning through prompts, questions, and motivational language.
• Innovations:
  • Designed a cross-age VUI health learning platform to explore how VUI can enhance learning outcomes through SRL strategies.
  • Quantified the relationship between interaction fluency (e.g., conversational interruptions), learning outcomes, and technology acceptance for the first time.
• Implementation Steps and Key Technologies:
  1. Designed the VUI as a learning companion to provide users with health knowledge.
  2. Used the Google Dialogflow ES platform to enable natural language processing for user intent recognition.
  3. Conducted controlled experiments with three learning modes, randomly assigning participants to learn different topics.
  4. Evaluated the effectiveness by quantifying interaction fluency and learning outcomes.

Research Findings

• Specific Findings:
  • Both younger and older adults achieved comparable learning outcomes using "Health Buddy."
  • Younger adults benefited more from scaffolding building and conceptual diagramming modes, while older adults showed no significant differences across the three modes.
  • Interaction fluency (i.e., the number of conversational interruptions) significantly impacted learning outcomes, regardless of age.
  • Older adults' acceptance of VUI was strongly influenced by interaction fluency, whereas younger adults were less sensitive to fluency issues.
• Comparison with Existing Solutions:
  • Most existing learning tools rely on graphical user interfaces (GUI), which present high learning barriers and complexity for older adults. This study demonstrates that VUIs are more accessible and user-friendly for older adults due to their simplicity in interaction.
  • "Health Buddy" showcased the feasibility of VUIs in informal learning scenarios, highlighting their intuitive and controllable nature compared to traditional internet searches.
• Evaluation and Experimental Results:
  • Older adults experienced an average of 1.36 conversational interruptions, while younger adults experienced 1.17 interruptions, with no significant difference.
  • Interaction fluency significantly influenced overall learning scores, particularly in scaffolding building and conceptual diagramming modes.
• Limitations and Future Directions:
  • Limitations:
    1. The experiment was conducted online due to the pandemic, potentially affecting sample diversity and representativeness.
    2. Most participants were highly educated native English speakers, limiting the generalizability of the findings.
  • Future Directions:
    1. Explore the applicability of broader learning strategies across different age groups, especially older adults.
    2. Conduct long-term studies in natural environments to observe behavioral adaptation and practical impacts.
    3. Develop error recovery mechanisms for VUIs to enhance interaction fluency and user acceptance.

Through this structured summary, the study demonstrates how VUIs can promote cross-age health information learning through self-regulated learning strategies, emphasizing the importance of factors such as interaction fluency. It also provides clear guidance for future designs that are friendly to older adults.