A Socially Relevant Artificial Intelligence Curriculum for Female High School Students

• In Event: Teaching Artificial Intelligence in K–12: Innovative Learning Designs in an Evolving Landscape

Abstract

OBJECTIVE
Here we describe the Socially Relevant AI and Machine Learning (ML) module designed with the goal of raising interest in advanced CS topics among high school students, and especially females (who currently comprise only 19% of CS undergraduates). We investigate how this curriculum affects students' self-efficacy and confidence in computing, especially for high school female students.

CURRICULUM CURATION
We piloted the AI/ML module in a series of summer camps using both block-based and text-based languages, Netsblox (Broll et al, 2017) and Python respectively. The camp presented a condensed version of the planned 9-week module which utilizes project based learning (Blumenfeld et al., 1991), pair programming (Werner et al., 2004), and societal implication discussions (Khan et al., 2016). For their final project, students are given agency to choose (Lytle et al., 2019) what themes to apply their newly understood AI or machine learning algorithms, such as Twitter crowdsourcing, lyric sentiment analysis, or bias identification in news articles. The camp and curriculum were intentionally designed to present representations of diverse women in computing, through female camp instructors from diverse race/ethnicity groups, the inclusion of women in images and videos, and women invited speakers. Furthermore, student group work was structured to ensure that no female students were the single female in a group.

METHODS & DATA
The AI/ML curriculum was piloted in a virtual, 2 week long summer camp for high school students with approximately 35 hours of instructional time. Of the 25 students who participated in the summer camp, 10 students identified as female and 15 identified as male (Table 2.1). Using a statistical software package, SPSS, an independent samples t-test was conducted on the pre and post camp surveys to compare female and male participant confidence before and after attending the 2021 summer camps. The pre and post survey entails 52 Likert type questions in the following subcategories: confidence, perception, interest in computing, interest in computing careers, and curriculum content knowledge.

FINDINGS
After piloting the materials, we found that there was a significant positive increase in female participant confidence and self efficacy in confidence in AI and ML content, self efficacy in computer science, and career identity after attending the AI/ML camp. Compared to male students, we found that females exhibited statistically significantly stronger increases in self efficacy in computing, confidence in communicating with computer scientists, and computing content (Table 2.2). We believe that the differences in outcomes for female participants were heavily influenced by our curriculum’s focus on human-centered computing, rather than a more traditional programming and syntax focus.
SIGNIFICANCE
By engaging with AI through socially relevant applications, participants were able to focus on AI concepts (rather than just coding) and engage in discussions focused on real-world applications. The results suggest that presenting AI and ML curricula with a socially-relevant focus, grouping female students with other females, and purposefully including women in the curricular materials and as instructors, can enhance female student experiences.

Authors

• Isabella Gransbury, North Carolina State University
  Presenting Author

• Veronica Catete, North Carolina State University
  Non-Presenting Author

• Tiffany Barnes, North Carolina State University
  Non-Presenting Author

• Shuchi Grover, Looking Glass Ventures
  Non-Presenting Author

• Brian Broll, Vanderbilt University
  Non-Presenting Author

• Akos Ledeczi, Vanderbilt University
  Non-Presenting Author