Non-verbal Collaborative Learning in Chess
Nov 22, 2024
Research @ Human-AI Integration Lab, UCSB
Supervised by Misha Sra & Arthur Caetano
Research Journey #1
As we have researched into the role of AI as peer in guiding human' learning in collaborative settings and motor learning environment, we have generalized the pros and cons of AI as peer, moderator, and instructor.
We found one particular interesting cons of AI Peer: Lack of non-verbal interaction.
As we noticed that AI lacks the non-verbal interactions as we have between human's peer learning, we want to investigate how non-verbal interaction between human and AI may affect how we collaborate and learning from AI.
Research Journey #2
Our research journey began with a shared curiosity about the evolving dynamics of human-AI interaction, particularly in scenarios where non-verbal communication and implicit collaboration are essential. To ground our investigation, we drew inspiration from a unique chess strategy shared by one of our team members, who serves as the president of the Chess Club at UCSB. He introduced us to a fascinating chess variant known as "Hand-and-Brain," where two players collaborate to play a single game:
One player (the "Brain") selects the type of piece to move (e.g., knight, rook, or pawn), while the other player (the "Hand") decides where to move it.
This interaction creates a compelling combination of strategic guidance and tactical execution, highlighting the nuances of non-verbal communication and coordination.
This real-world example provided a powerful analogy for understanding the potential of human-AI collaboration. Like the Hand-and-Brain team, humans and AI can bring complementary strengths to the table: humans contribute strategic reasoning and contextual understanding, while AI offers computational speed, pattern recognition, and analytical rigor. We envisioned scenarios where non-verbal cues, shared intentions, and trust-building mechanisms could elevate such collaborations, moving beyond traditional input-output models toward more seamless and intuitive interactions.
My Role
I contributed to the project by focusing on the following areas:
Literature Research: Conducted a thorough review of prior studies on the dynamics of human-AI interaction in chess, particularly in collaborative settings like Hand-and-Brain, to inform our theoretical framework.
Experimental Design: Assisted in brainstorming the experimental setup, discussing potential pilot studies, and identifying necessary preparations, such as technical equipment and data collection protocols.
Data Analysis: Conducted analysis on key data streams, including eye-tracking patterns, facial expressions, and chess move logs, to uncover insights into non-verbal communication and collaboration dynamics.
Python Development: Developed Python scripts to fetch participants’ chess ratings from online databases and designed a matching algorithm to pair participants with similar ratings for balanced gameplay in the study.