ShamAIn: Designing Superior Conversational AI Inspired by Shamanism
Honorable MentionAuthors
Research Background and Issues
What problems or challenges did the authors identify?
While the application of Conversational Artificial Intelligence (CAI) has rapidly expanded across various domains such as healthcare, education, and mental health, current research primarily focuses on enhancing functionality and user experience, treating CAI mainly as a tool or collaborator. However, few studies have explored designing CAI as an entity perceived as superior to humans, capable of providing extraordinary insights.
Why is this issue important?
- Potential for new interaction possibilities: Designing CAI as a superior entity could deepen the relationship between users and the system, fostering more open communication, especially in contexts involving personal issues and emotional support.
- Cultural and psychological factors: In certain cultural contexts, particularly those with spiritual belief systems, perceiving technology as a supernatural or mystical force may influence its acceptance.
- Redefining the role of CAI: Exploring the possibility of CAI transcending its role as a tool could introduce new dimensions to human-computer interaction.
Research Motivation and Related Work
The study draws inspiration from Korean shamanistic traditions, where shamans (Mudang) are often regarded as intermediaries capable of communicating with supernatural entities. The research builds on the following trends:
- People tend to develop "magical thinking" about complex technologies, especially when their behavior is difficult to explain.
- Religious and spiritual beliefs often influence the deployment and design of technology (e.g., Buddhist rituals in China, family altars in Japan). Thus, the authors aim to incorporate symbolic elements of Korean shamanism into the design of CAI, enabling it to be perceived as possessing a "superhuman" quality.
Solution
What methods or solutions did the authors propose?
- Designing the ShamAIn system: Develop a multimodal conversational AI system that integrates the philosophy and symbolic design of Korean shamanism.
- Using a Research through Design (RtD) approach: Conduct interviews with users of Korean shamanism to identify design elements that evoke a sense of "transcendence" and incorporate these elements into ShamAIn's interaction design.
What are the innovative aspects of this solution?
- Culturally driven design: By centering the design on the symbolic system of Korean shamanism, the study combines technology with cultural traditions to offer a unique interaction experience.
- Perception of transcendence: The study moves beyond the conventional tool-centric design of CAI to explore innovative directions where the system is perceived as having "spirituality."
- Perception of intangible presence: The research breaks away from the tangible and concrete framework of CAI, attempting to create an interaction atmosphere that feels "intangible yet present" through multimodal elements like sound, visuals, and scent.
What are the implementation steps and key technologies used?
- Design Strategies: Based on interview data, the authors identified four key design strategies:
- Create an "extraordinary" atmosphere.
- Evoke a sense of awe or even fear.
- Convey a feeling of "intangible presence."
- Make users feel that the AI "knows" personal information about them.
- Designing the Environment and Interaction Modes:
- Environment Design: Set up a dark and mysterious interaction booth (including scents, LED lights, decorative paintings, etc.).
- Interaction Modes: Utilize multisensory interactions (e.g., chimes, background music, vocal tone).
- Building the Voice Interaction System:
- Use GPT-4o to generate conversational content, converting it into voice output via Text-to-Speech (TTS) technology.
- Design guiding prompts to ensure the conversational content is authoritative and personalized.
- Implementation and Exhibition: Install the interactive platform in a university exhibition hall and conduct a six-week experimental test.
Research Findings
What specific results were achieved?
- Perception of transcendence and reverence:
- Participants reported a strong sense of awe towards ShamAIn, perceiving it as having supernatural qualities.
- The mysterious atmosphere and interactive details, such as chimes, lighting, and scents, enhanced this perception.
- Open communication:
- The experiment showed that participants gradually disclosed deeper personal issues over multiple interactions and were more willing to accept ShamAIn's advice.
- Compared to traditional CAI, participants were more inclined to trust ShamAIn's responses.
What advantages does it have over existing solutions?
- Depth of emotional interaction: Designing CAI as a transcendent entity effectively breaks the limitations of traditional tool-based interactions, encouraging users to open up emotionally.
- Cultural integration: The design strategy based on Korean shamanism provides a highly localized interaction experience, creating a distinct identity compared to conventional CAI.
- Multimodal interaction details: This study employs multisensory integration, offering a new approach to creating metaphorical technological designs.
What were the experimental or evaluation results?
- A total of 20 participants took part in the experiment, with over 90 visits recorded (an average of 4.5 visits per participant).
- Participant feedback indicated a high level of trust in ShamAIn, with users increasingly willing to rely on its advice.
- The system's interaction content and tone were perceived as shaman-like, altering participants' traditional perceptions of AI.
Limitations and Future Directions
Limitations:
- Cultural adaptability: This study is primarily based on the Korean cultural context, and its adaptability to other cultures requires further validation.
- Long-term effects: The long-term psychological dependence and social impact of ShamAIn on users remain unclear.
- Privacy and security issues: The disclosure of personal information by users may pose risks of privacy breaches.
Future Directions:
- Exploring multicultural designs: Extend the shamanistic interaction model to other cultural contexts (e.g., tarot cards, Western astrology).
- Applications in psychological support: Investigate the practical applications of this design framework in mental health support or crisis intervention.
- Optimizing privacy management: Develop more secure mechanisms for protecting user data to enhance the system's privacy and security.
Through this study, the authors have opened a new frontier for conversational AI and provided important insights into deeply exploring the potential for intelligent agents to transcend their role as mere tools.
Research Questions / Practical Problems
Question signals indexed for this paper.
Research Questions
3- How can LLM-based educational agents be designed to support collaborative learning among middle school students?Category: Trust in Conversational AI and ChatbotsSimilar questionsarrow_forward
- Which dialogue strategies and techniques can effectively promote knowledge sharing and deep interaction among students?Category: Trust in Conversational AI and ChatbotsSimilar questionsarrow_forward
- How can automatic AI intervention and human expert oversight be balanced in educational technology?Category: Trust in Conversational AI and ChatbotsSimilar questionsarrow_forward
Practical Problems
1- Teachers struggle to monitor student collaboration in real time, leading to insufficient collaborative skills training.Category: Trust in Conversational AI and ChatbotsSimilar questionsarrow_forward
- 80%
LLooM: Weaving Stories and Probing Experiences of Language Technologies
CHI '26· Conversational Chatbots +2
- 60%
The Effect of Co-located Audiences on User Experience with Conversational Interfaces in Physical Spaces
CHI '19· Conversational Chatbots +1
- 60%
Chatbots, Humbots, and the Quest for Artificial General Intelligence
CHI '19· Conversational Chatbots +1
- 60%
Synlogue with Aizuchi-bot: Investigating the Co-Adaptive and Open-Ended Interaction Paradigm
CHI '24· Conversational Chatbots +1
- 60%
Apple’s Knowledge Navigator: Why Doesn’t that Conversational Agent Exist Yet?
CHI '24· Conversational Chatbots +1
- 60%
Chatbots for Data Collection in Surveys: A Comparison of Four Theory-Based Interview Probes
CHI '25· Conversational Chatbots +1
- 60%
When Personas Talk to You: Evaluating the Evolution of User Personas from Static Profiles to Conversational User Interfaces
DIS '25· Conversational Chatbots +1
- 60%
BigBlueBot: Teaching Strategies for Successful Human-Agent Interactions
IUI '19· Conversational Chatbots +1
Based on Jaccard similarity of research subtopics & professions (≥60%)