Abstract
As Artificial Intelligence (AI) increasingly permeates various sectors of society, the need for comprehensive AI literacy in K-12 education grows. However, AI education is often restricted to Computer Science (CS) departments, leaving teachers in humanities and other STEM fields without adequate resources. This paper presents a co-design study with eight high school teachers from diverse disciplines to develop modular AI curricula. We introduce the "dual exploration" framework, where students use disciplinary tools to critique AI while leveraging AI to deepen subject-matter understanding.
Introduction
The rapid advancement of Generative AI and Large Language Models (LLMs) has sparked significant interest among educators. While CS teachers focus on the technical mechanics of these systems, teachers in English, Social Studies, and Art are grappling with how these tools reshape their fields. Current AI curricula often fail to bridge this gap, focusing either on coding or abstract ethics without connecting to specific disciplinary pedagogies. This study explores how to empower non-CS teachers to integrate AI into their existing classrooms through a collaborative co-design process.
Methodology and Participants
The researchers utilized a dual-purpose co-design model rooted in Research-Practice Partnerships. This approach involved five collaborative sessions with eight high school teachers. These sessions moved from building foundational AI literacy to outlining specific lesson plans tailored to each teacher's subject area.
The participant group was intentionally diverse, representing various school types (public, charter, technical) and serving different student demographics, as shown in the table below.
The sessions were recorded and transcribed, totaling 13 hours of discourse. Thematic analysis was then applied to these transcripts and the artifacts created by teachers (e.g., lesson outlines and digital whiteboards) to identify key integration patterns.
The "Dual Exploration" Framework
A central contribution of this work is the "dual exploration" framework. This concept describes a reciprocal relationship between AI tools and disciplinary knowledge:
- Disciplinary Frameworks for AI Critique: Students apply established analytical tools from their subjects to evaluate AI. For example, using the SOAPSTone framework in English to analyze the speaker, occasion, and purpose of an AI-generated text, or using OPCVL (Origin, Purpose, Content, Value, and Limitation) in History to assess AI-generated historical narratives.
- AI Tools for Disciplinary Depth: Conversely, students use AI to explore the boundaries of a subject. An art student might use DALL-E’s anatomical errors to spark a discussion on human artistic precision and the nuances of physical form.
Key Insights
The Technical-Ethical Dichotomy
The study revealed a distinct split in how teachers conceptualized AI based on their background. Engineering and Math teachers were primarily interested in the "technical guts"—the inputs, outputs, and mathematical underpinnings like mean squared error. In contrast, Humanities teachers prioritized "making an argument," focusing on debating social impacts and the ethical implications of AI deployment.
AI as an Equalizer and the "Breaking Point"
Teachers viewed AI as a potential "equalizer," allowing students who might lack specific technical skills (e.g., drawing or complex coding) to engage in high-level creative and conceptual work. However, they also noted that students often try to "break" the tools. Rather than seeing this as a distraction, teachers proposed using these "breaking points"—such as hallucinations or rendering failures—as pedagogical opportunities to explain the underlying limitations of AI models.
Practical Implications for Curriculum Design
The co-design process highlighted several requirements for successful AI integration:
- Modularity: Given the rigid and varied nature of high school schedules, teachers preferred modular resources that could be "plugged into" existing units rather than a standalone, full-semester curriculum.
- Practicality: Teachers, particularly in English Language Arts (ELA), found practical skill-building (e.g., using evidence to critique a specific tool's failure) more effective than abstract ethical discussions.
- Professional Agency: The study underscored that even experienced CS teachers feel gaps in their knowledge regarding rapidly evolving Generative AI. This suggests a need for ongoing, collaborative professional development that respects teacher expertise.
Conclusion and Future Work
This research demonstrates that AI education can and should extend beyond the CS classroom. By leveraging the "dual exploration" framework, educators can create meaningful connections between AI and diverse disciplines. Future work should focus on expanding this model to more diverse cultural contexts and developing "micro/macro" ethics frameworks that bridge technical details with broader societal impacts. Decoupling AI from traditional programming may also lower the barrier to entry for non-CS educators.
Summary updated