AI4GA

The AI4GA project is developing a nine-week AI elective course for middle school students in Georgia called Living and Working with Artificial Intelligence. Researchers co-designed curriculum and assessments with computer science teachers from school districts serving three ethnically and geologically diverse populations in Georgia: urban, suburban and rural. Since the spring semester of 2022, 20 teachers offered the curriculum to a total of 1459 students. 

Our current research involves the analysis of Unit 1’s MyDreamBot project. This project asked students to imagine and design an AI-powered robot meant to solve a societal problem that they view as important. Students created their robots via a range of formats, including drawings, 3D models via TinkerCAD and Hero Forge.  

Below are some of our past research products: 

This paper investigated the efficacy of formative AI knowledge assessment instruments and was submitted to the International Computing Education Conference (ICER) in 2024. We found that despite some struggles with a few topics (e.g., breadth-first search algorithms), students developed an understanding of sensor functions and their limitations, and can critically evaluate the societal impacts of autonomous vehicles. Design considerations for formative assessments were proposed. For example, assessments should strike a balance between structured elements and room for creativity to ensure that the assessed knowledge is surfaced in open-ended projects. 

This paper investigated the appropriate format for middle school AI curricula and was submitted to the International Journal of Artificial Intelligence in Education (IJAIED) in 2023. We found that the use of real-world examples of AI shown in the classroom to be effective to drive engagement and student interest. Students also were shown to be highly engaged when allowed to participate in collaborative activities such as debates, discussions around AI ethics, both in the classroom and through interactive platforms such as NearPod or PearDeck.

This paper investigated how middle school students engaged with technical content about AI. It was accepted by the ACM Special Interest Group on Computer Science Education (SIGCSE) in 2023. Findings indicate that post-COVID, students were comfortable working on individual assignments at their own pace. They also preferred hands-on unplugged activities (e.g., using graph coloring to solve Georgia highway route-finding tasks) over knowledge-focused instruction due to the nature of elective courses. 

This poster described how we used co-design both as a tool for curriculum development and as a mechanism for mutual professional development with middle school computer science teachers. It was accepted by the the ACM SIGCSE in 2022. Across three phases of co-design, teachers were able to organically adapt the AI curriculum to their student interests and needs. We also increased understanding of what support teachers need to confidently teach AI and how to scaffold students’ AI learning.