Shaping AI Interest in Rural Middle Schools with Unplugged Learning: Gender Differences and Teacher Insights

Hansol Lim, Danielle Boulden, Jessica Vandenberg, Veronica Cateté, Wookhee Min, Bradford W. MottAAAI 2025 · DOI

TL;DR

This paper introduces "Guess Whose Face," an unplugged, game-based learning activity designed to teach foundational AI concepts (Representation & Reasoning) to rural middle school students. The study found that the game significantly boosted AI attitudes in female students and highlighted the importance of teacher training and support for effective implementation in under-resourced communities.

What Problem Does This Paper Address?

The rapid integration of Artificial Intelligence (AI) into daily life necessitates widespread AI literacy. However, access to AI education remains uneven, particularly in rural communities. This disparity is exacerbated by a lack of formal AI curricula and a shortage of trained educators. The paper addresses this gap by developing and evaluating an "unplugged" (non-computer-based) learning activity to introduce AI concepts to middle schoolers in rural North Carolina, aiming to spark interest and understanding, especially among demographics historically underrepresented in STEM.

Methods

The core of the study is the "Guess Whose Face" game, an unplugged activity designed to teach the AI concept of "Representation and Reasoning." The game is a card-based guessing game where one player (the drawer) uses facial features (represented by dots on tracing paper) to help another player (the guesser) identify a hidden character. The game involves elements of feature extraction and noise, mirroring real-world AI processes.

The study employed a mixed-methods approach:

  • Game Design & Iteration: The game was iteratively refined through pilot testing.
  • Professional Development (PD): Middle school teachers received a day-long PD session to learn about AI concepts and the game, including hands-on play and access to instructional materials.
  • Classroom Implementation: Two teachers implemented the game in their classrooms with approximately 90 students.
  • Student Surveys: Pre- and post-surveys measured student attitudes towards AI and conceptual understanding using Likert scales and multiple-choice questions.
  • Teacher Interviews: Post-implementation interviews gathered teacher insights on facilitators, challenges, and student engagement.
  • Statistical Analysis: T-tests and ANOVA were used to analyze survey data, calculating normalized learning gains and effect sizes.

Datasets

Not discussed in the paper. The study utilized survey data collected from participating students and qualitative data from teacher interviews.

Key Insights and Ideas

  • Unplugged Learning as an AI Gateway: "Guess Whose Face" demonstrates that engaging, hands-on, unplugged activities can effectively introduce complex AI concepts like Representation and Reasoning to middle schoolers, even in resource-limited settings.
  • Gendered Impact on Attitudes: The game had a statistically significant positive impact on female students' attitudes towards AI, suggesting it can help bridge gender gaps in STEM interest. While male students also showed a slight increase, it wasn't statistically significant.
  • Teacher Scaffolding is Crucial: Teachers emphasized the vital role of the PD session, hands-on practice, and instructional slides in building their confidence and ability to implement the AI activity effectively.
  • Time Constraints are a Major Hurdle: The limited class time (40-45 minutes) prevented students from fully engaging with all game levels, hindering deeper exploration of AI concepts.
  • Curiosity Beyond the Game: The activity sparked student curiosity about AI, leading to discussions on its broader societal implications, suggesting it can serve as a springboard for further AI learning.

Key Results

  • Overall Attitudes: Student attitudes towards AI increased slightly from a pre-survey mean of 11.15 to a post-survey mean of 11.58, though this was not statistically significant (p = .38).
  • Overall Knowledge: Knowledge scores showed a marginal trend towards statistical significance, increasing from 2.01 to 2.18 (p = 0.09).
  • Gender Differences (Attitudes): Female students showed a statistically significant increase in attitude scores (p = 0.02) with a moderate effect size (0.39), while male students' attitude changes were not statistically significant.
  • Gender Differences (Knowledge): Neither male nor female students showed statistically significant changes in conceptual understanding scores.
  • Grade Level Differences (Attitudes): Grade 7 students showed a significant increase in attitude scores (p = 0.03) with a moderate effect size (0.46). Grade 6 and 8 showed non-significant increases.
  • Grade Level Differences (Knowledge): No grade level showed statistically significant changes in conceptual understanding.

Limitations and Future Work

Limitations: * Small Sample Size: The study had a limited number of participants, especially considering students who did not complete the post-survey. * Non-Representative Sample: Participants were from two schools within the same system and all received gifted services, limiting generalizability to the broader middle-grade population. * Brief Intervention: The short duration of the activity and surveys may have reduced the likelihood of observing significant changes in student attitudes or learning.

Future Work: * Longer Studies: Conducting longer studies with more rounds of gameplay at various difficulty levels is recommended to better observe changes in conceptual understanding. * Broader Implementation: Expanding the study to include a larger and more diverse student population is crucial. * Extension Activities: Developing further unplugged and plugged activities, along with teacher prompts, to explore AI concepts and their ethical implications more deeply. * Curriculum Integration: Creating additional resources for teachers to integrate AI education, including its ethical dimensions, into K-12 classrooms.

Summary updated

BibTeX
@inproceedings{lim2025shaping,
  author = {Hansol Lim and Danielle Boulden and Jessica Vandenberg and Veronica Cateté and Wookhee Min and Bradford W. Mott},
  title = {{Shaping AI Interest in Rural Middle Schools with Unplugged Learning: Gender Differences and Teacher Insights}},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year = {2025},
  doi = {10.1609/aaai.v39i28.35188},
}
Figure