Project Overview

This project involves developing upper elementary school students’ abilities to work with artificial intelligence (AI) and in future careers. Through this project, a team of robotics and education researchers at Tufts University in Massachusetts and Maryville University in St. Louis, MO is working with teachers in St. Louis County to develop a research-informed educational ecosystem introducing AI, specifically machine learning (ML), concepts to upper elementary school students. (Photo Credit: Milan Dahal)

Objectives

1. To create, test, and revise inexpensive hardware with multiple built-in ML algorithms to allow students to explore the learning behavior of different algorithms.
2. To create a system that is highly compatible with 5th grade classrooms and specifically focus on hardware that is safe, inexpensive, and easy for 5th graders to use and learn from and with.
3. To discover through student and teacher feedback which ML algorithms are most powerful and which are easiest to use.

Research Questions

1. How does the introduction of tangible artificial intelligence elements lead to changes in upper elementary students’ understanding of artificial intelligence concepts and attitudes towards artificial intelligence?
2. How do different levels of complexity and variety of tangible artificial intelligence elements impact the engagement of upper elementary students and the diversity of their solutions and designs?
3. What are the potential benefits and challenges of introducing tangible artificial intelligence elements in integrated engineering and literacy activities?

Key Innovations

Smart Motor Platform

Milan Dahal, graduate student, developed the “Smart Motor” embedded platform, a low-cost hardware and software toolkit that enables students to train simple machine learning, nearest neighbor, models through a built-in tangible interface. This work included:

• Computer-free interface to keep classroom costs low
• Engaging and interactive code-free training of motor behaviors for integrated engineering activities
• Usability testing of the on-motor interface

The open-source designs are available on the project website.

Visual Machine Learning Interface

Tanushree Burman, graduate student, developed the “Smart App”. An on-computer supplemental interface for the Smart Motor tool. This visual programming environment makes machine learning concepts accessible by:

• Representing sensor, output and data representation through intuitive graphical elements
• Visualizations of classification decision-making processes
• Providing immediate feedback during model training and testing
• Supporting math standards interpreting data and graphs

Additionally, this work has been extending into the development of interfaces for helping elementary age students also understand additional machine learning methods including reinforcement learning.

Co-Design Workshop with Teachers

Geling (Jazz) Xu, graduate student, developed and led co-design activities with our partner teachers. During these workshops the partner teachers explored the Smart Motor and Smart App platforms, and working collaborative on the design of classroom activities for introducing these tools to students. These workshop have helped to inform both machine learning instructional design for elementary students as well as informing the development of experiences for introducing educators to machine learning and AI technologies.

Contribution

As principal investigator, I provided support and direction to the interdisciplinary and multi-organization team of computer scientists, engineers, educators, and researchers. I provided expertise and advising at the intersection of interaction research, emerging technologies, and instructional design. I evaluated progress toward objectives through regular meetings, management of the timeline and budget, and reporting to the funding agency.

Related Publications