Using Embodied Knowledge to Design, Augment, and Evaluate Machine Learning Models

• In Event: Bridging the AI Literacy Gap: Systems-Level Approaches to K-12 CS Education

Abstract

Abstract: This study explores how learners use embodied knowledge to design, augment, and evaluate machine learning models. Fifteen middle school girls completed a curriculum designed to blend machine learning concepts with artificial intelligence literacy. Analysis of girls’ models, interactions, and post-interviews revealed patterns in how participants use a variety of embodied resources in each step of their model development. The findings reveal a framework of collaborative embodied engagement that illustrates how embodied knowledge operates across three interconnected phases of ML model development: design (drawing on funds of knowledge), augmentation (using micromovements for data variation), and evaluation (collaborative testing and iteration).

Authors

• Alicia Lane, Vanderbilt University

• Atefeh Behboudi, Vanderbilt University

• Omer Zahid, Vanderbilt University

• Ashley Markatos, Vanderbilt University

• Golnaz Arastoopour Irgens, Vanderbilt University