1 of 18

This work is supported by NSF DRL-1949384 and DRL-DRL–1949493

ImageSTEAM: Teacher Professional Development for Integrating Visual Computing into Middle School Lessons

Suren Jayasuriya, Kimberlee Swisher, Joshua Rego, Sreenithy Chandran, John Mativo, Terri Kurz, Cerenity Collins, Dawn Robinson, Ramana Pidaparti

2 of 18

Full Team across ASU and UGA

Sreenithy Chandran

3 of 18

Project overview

Why?

Project Mission: Increase knowledge, motivation and interest of Middle School students in STEM careers in AI and computer vision

What?

Central Challenge: How do we incorporate AI into the middle school curriculum in public education? How can teachers of all disciplines be empowered to teach this material?

How?

Summer Professional Development for Teachers

Co-create learning modules with AI researchers

imageSTEAM.org

4 of 18

Project topics

Cameras

Images and pixels

Digital Colors

Image Processing

Data

Computer Vision

Machine Learning

5 of 18

Week 1

Week 2

Week 3

Week 0

Program structure

Schedule

(Arizona program site details)

introduction to program topics AI and computer vision

Mornings: Teachers observe class, “How Computers See” --->

Afternoons: work with researchers to develop a new lesson

Mornings: Teachers teach their lesson to students

Afternoons:more co-development and revisions

Mornings: Teach lesson to new students

Afternoons: Documentation with written lesson plans and videos

6 of 18

By The Numbers

Since 2021-2023

2 locations

Tempe, AZ

Athens, GA

43 teachers

In-Service Teachers, middle school level, all different disciplines

26 videos

Short-form videos about AI, visual media and computer vision for an audience of teachers and students

189 students

Students in grades 6-8 in Arizona and Georgia

7 of 18

Project Curriculum

1. Standards-aligned

To state or national standards

2. Flexible

“a la carte” if possible

3. Accessible

Free, easy to use software

4. Minimize need for background knowledge

Provide clear and direct support for teachers

5. Co-developed

Researchers provide disciplinary resources; Teachers provide expert instructional design and standards-alignment

Our rules for lesson development

imageSTEAM.org

Say as one sentence!!

We tried to constrained our curriculum to a few rules:

most lessons could be used one-off, not necessarily as part of a longer unit.
All lessons needed to be standards-aligned, meaning, the lessons needed to address a standard the teacher already has to teach.
Lesson content would require only a little bit of background knowledge, and lesson materials such as videos and slides would contain all the more complex explanations necessary so that teachers wouldn’t need to be super familiar with all the details, but could still facilitate the lesson.
We believe to do this well requires a broad scope of knowledge, that can’t be done alone. We need to leveraging teacher’s expertise in curriculum, standards, and student ability and interest to co-create lessons that could actually be adopted in public schools.

AI Researchers created several short-form video lessons on AI concepts to present as part of the PD program.

Get ready for research!

8 of 18

Project Curriculum

Free, easy-to-use tools online

imageSTEAM.org

9 of 18

Project Curriculum

We also developed custom notebooks in Google Colab for teachers and students to run

Created an “ImageSTEAM” package in Python that made coding for these notebooks easier (will be released by Summer 2024)

Caveat:

Difficulty level of these notebooks and Python coding

Some custom development

imageSTEAM.org

10 of 18

Sample Lessons

Sonar, Plate Tectonics, and Machine Learning

Natalie Carpenter

imageSTEAM.org

11 of 18

Sample Lessons

Visualizing Literature Through AI Text-to-Image

Sam Young

imageSTEAM.org

12 of 18

Sample Lessons

Plant Vs Animal Cell Diagrams

imageSTEAM.org

13 of 18

Behind The Scenes Videos

imageSTEAM.org

14 of 18

Research Conclusions and Future Work

15 of 18

Current Project Conclusions and Limitations

Co-designed curricular model is successful way of developing useful curriculum while educating teachers about the scope of the field

Humanities and arts classes are good for introducing concepts in AI and computer vision, and helped kids think about AI from an ethical, cultural perspective.

Limitations:

Teachers felt apprehensive about teaching lessons on less familiar concepts and using new tools

Technology access in public schools

AI generated image, Adobe Firefly Model 2

16 of 18

Research on Teachers

Leveraged Personal Construct Theory (Kelly, 1955) to analyze teacher perceptions about content material

Results showed positive changes in thinking about concepts such as computer vision and artificial intelligence highlighted by changes in dendrogram structure and clustering

Research still on-going for questions about student learning and STEM identities

17 of 18

Next Steps

On-going project through 2025

Animations to supplement instructional material

Summer 2024: 1-week workshops with 45 teacher slots

Published and accessible Colab Notebooks as well as accompanying Python library via GitHub and other public means

Storyboard for Neural Network animation

18 of 18

Thank you!

This work is supported by NSF DRL-1949384 and DRL-DRL–1949493

Photo by Joshua Fuller on Unsplash

Published curriculum is freely available:

ImageSTEAM.org/co-designed-curriculum

ImageSTEAM.org/videos

kimberlee.swisher@asu.edu

sjayasur@asu.edu

Photo by Joshua Fuller on Unsplash

https://www.imagesteam.org/videos

Who am I

Overview of program 1 sentence

Title of proceedings -- Terri Kurz

Details of ImageSTEAM justification, goals, what is AI.

Structure of the PD program, pros and cons. DCSI

Findings, that we can change perception of AI, that we can create rich curriculum but that integrating into classrooms has so many challenges (what are these challenges, that humanities and english classes have more flexibility than science and math and that these subjects might work better anyway as we want kids thinkning about AI from an ethical holistic perspective, not just HOW can we do, but why, and what we might not want to do, and how humans interact with technology and technology changes us.

Website and Curriculum details, including video process.