Can a Humorous Conversational Agent Enhance Learning Experience and Outcomes?

Honorable Mention
Conversational ChatbotsAgent Personality & AnthropomorphismIntelligent Tutoring Systems & Learning AnalyticsUniversity Professors & ResearchersSpecial Education Teachers

Title of the Paper

Can a Humorous Conversational Agent Enhance Learning Experience and Outcomes?

Paper Information

  • Subject Area: Educational Technology and Human-Computer Interaction (Role of Humorous Conversational Agents in Education)
  • Keywords: Conversational Agent, Education, Motivation, Learning Experience, Learning Outcomes, Humour, Human-Computer Interaction, Pedagogical Agents, Artificial Intelligence, Learning-by-Teaching

Research Background and Questions

  • Research Background:

    • Conversational agents have become increasingly prevalent in computer-assisted learning environments, with research aiming to understand how to design these agents to optimize learning experiences and outcomes.
    • The use of humor in learning has been shown to reduce anxiety, boost self-esteem, and enhance motivation, yet systematic studies on humor in teaching conversational agents remain scarce.
    • Special focus is placed on how the humor style of "Teachable Agents" influences users' perceptions, attitudes, and teaching behaviors.
  • Research Questions:

    1. How do different humor styles (affiliative humor and self-deprecating humor) in teachable agents affect participants' perceptions of the agent and teaching habits?
    2. Do learners' personality traits (e.g., humor style) interact with the agent's humor style to further influence learning experiences and outcomes?
  • Research Motivation:

    • Humor has been proven effective in fostering social bonds, alleviating anxiety, and increasing engagement in traditional learning contexts, but its role in conversational agents has yet to be systematically explored.
    • Integrating humor may enhance the "human-like" quality and engagement of conversational agents, thereby influencing learning behaviors and experiences.

Solution

  • Research Methods:

    • Using the Curiosity Notebook platform, the experimental design compares three conditions (affiliative humor, self-deprecating humor, no humor).
    • The system records user-agent interactions and employs questionnaires to quantify participants' perceptions of the agent, learning motivation, and emotional states, analyzing the multi-level impact of humor on learning.
  • Innovation and Key Points:

    1. Systematic exploration of humor styles in teachable agents, addressing a gap in academic research.
    2. Advanced experimental design and multi-dimensional measurements (e.g., interaction data recording, neuropsychological scales) to evaluate humor effects.
    3. Consideration of individual characteristics (participants' humor styles) to investigate the "fit between personality and interaction design."
  • Implementation Steps and Techniques:

    1. Using the Curiosity Notebook platform, users interact with the agent through teaching tasks.
    2. The humor styles of the agents are specified: affiliative humor characterized by harmony and jokes, self-deprecating humor characterized by expressions of personal inadequacy.
    3. Collect participants' behavioral data during teaching tasks, such as keystrokes, article visits, and response content, while recording knowledge changes before and after learning and questionnaire-based perceptions and motivations.

Research Findings

  • Specific Findings:

    • Affiliative humor significantly increased participants' motivation and effort during tasks.
    • Self-deprecating humor moderately increased teaching effort but negatively impacted the experience, potentially reducing participants' enjoyment and confidence in teaching.
  • Advantages:

    • Provides a comprehensive perspective on the potential benefits and challenges of humor in teachable agent roles.
    • Further emphasizes the importance of "personalized design" matching user characteristics (humor style should align with learner traits).
  • Experimental and Evaluation Results:

    1. Humor perception: Agents in humorous conditions were perceived as having higher humor levels and were considered more "human-like" compared to non-humorous agents.
    2. Learning performance: No significant difference was found in knowledge transfer or participants' knowledge gains between humorous and non-humorous agents.
    3. Motivation and effort: Affiliative humor conditions significantly extended interaction time and enhanced extrinsic motivation, though frequent humor could distract participants.
    4. Individual interaction: Participants with a tendency for self-enhancing humor preferred affiliative humor agents and disliked self-deprecating humor agents, while those with aggressive humor styles engaged more actively with both humorous agents.
  • Limitations and Future Directions:

    1. Limitations:
      • Participants were adults (18-35 years old), limiting the generalizability of the results to broader age groups.
      • Focused on two types of verbal humor (affiliative and self-deprecating), leaving other humor genres (e.g., non-verbal humor) unexplored.
      • Knowledge measurement emphasized memory rather than deep understanding or transferability.
    2. Future Directions:
      • Investigate the optimal timing and frequency of humor usage.
      • Expand research to other learning topics and broader participant demographics.
      • Test combinations of multiple humor types to explore their potential impact on learning outcomes.

Conclusion

This study demonstrates that humor plays a significant role in educational conversational agents, but balancing different humor styles is crucial to enhancing or hindering the learning experience. Future designs should focus on individual learner characteristics, optimizing humor frequency and style combinations to improve sustained engagement and learning motivation in educational tasks.

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/papers/chi/47686/2021

AdRecommended

Learn AI Coding at CodeNow

open_in_newOpen DOI Link
DOI: https://doi.org/10.1145/3411764.3445068
At a Glance

Paper Snapshot

fact_check
dataset
Source
CHI
calendar_month
Year
2021
emoji_events
Award
Honorable Mention
group
Authors
5 authors
sell
Subtopics
Conversational Chatbots, Agent Personality & Anthropomorphism, Intelligent Tutoring Systems & Learning Analytics
work
Professions
University Professors & Researchers, Special Education Teachers
article
Content Status
Full text indexed
hub
Related Papers
0 related papers