Human-AI communication for human-human communication: Applying interpretable unsupervised anomaly detection to executive coaching
Project Overview
The document explores the role of generative AI in education, specifically focusing on its application in executive coaching through tools like REsCUE and INWARD, which utilize unsupervised anomaly detection to provide coaches with interpretable cues about nonverbal behaviors. These AI tools aid coaches in observing and interpreting social signals, thereby enhancing their understanding of clients' internal states. This methodology not only enriches the coaching process but also creates educational opportunities for novice coaches by encouraging reflective learning and discussions with experienced mentors. The authors highlight the significance of human-AI collaboration in navigating complex social interactions and propose that unsupervised techniques can significantly enhance AI's contributions to educational contexts and interpersonal communication. Overall, the findings suggest that generative AI can effectively support professional development in coaching by fostering deeper insights into client dynamics and improving the coaching experience.
Key Applications
Coaching Support with Nonverbal Analysis
Context: Executive coaching and training of novice coaches through video reflection sessions, where experienced coaches and coachees analyze nonverbal cues in recorded coaching sessions.
Implementation: Utilizes unsupervised anomaly detection to analyze nonverbal cues in coaching sessions. Coaches are provided with an interface to review these sessions, annotate the cues, and engage in discussions about their interpretations, promoting critical thinking and collaborative learning.
Outcomes: Improves coaches' ability to observe and interpret nonverbal behavior; facilitates learning opportunities for novice coaches; enhances interpretability and contextual understanding; encourages critical discussions about discrepancies in interpretations.
Challenges: May require validation across different coaching and educational contexts; potential for novice coaches to learn without computational tools; reliance on human interpretation may vary; authority bias may affect learning.
Implementation Barriers
Technical
Difficulty in relying on outputs from AI when they contradict expert intuition.
Proposed Solutions: Enhancing the interpretability of AI outputs to align with expert observations.
Educational
The need for intensive training in interpreting nonverbal cues for novice coaches.
Proposed Solutions: Using tools like INWARD and REsCUE to provide structured learning and reflection opportunities.
Project Team
Riku Arakawa
Researcher
Hiromu Yakura
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Riku Arakawa, Hiromu Yakura
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai