High school teachers can successfully integrate generative AI into their subject-area lessons through collaborative co-design approaches, creating meaningful learning experiences that position students as critical evaluators rather than passive consumers of AI technology.
Objective: The main goal of this study was to investigate how high school teachers design and implement lessons about generative artificial intelligence (GenAI) within their specific subject areas. The researchers aimed to understand what pedagogical approaches teachers use when integrating GenAI tools into their curricula and how professional learning communities can support teachers in developing effective AI-integrated lessons. The study specifically examined whether teachers would position GenAI as an "object-of-study" (teaching about AI as a technological tool) or as "subject-specific" (integrating AI within their disciplinary content), and how co-design experiences influence their pedagogical design capacity for GenAI instruction.
Methods: The researchers conducted descriptive case studies of two high school teachers, June (AP Research methods teacher) and Margot (AP English Language and Composition teacher), who participated in CRAFT (Classroom-Ready AI resources For Teachers), a year-long professional learning co-design partnership. The CRAFT program consisted of six 60-80 minute virtual sessions held from October to May, combining synchronous and asynchronous activities. The methodology included content analysis of lesson materials, thematic analysis of teacher interviews and reflections, and iterative development of analytic cases. Data sources encompassed professional learning session recordings, participant interviews, lesson design artifacts (written plans, Google Slides, activity guides), exit tickets, and teacher application materials. The researchers used Brown's pedagogical design capacity framework to analyze how teachers leveraged both internal resources (their subject knowledge and teaching experience) and external resources (the co-design community) to create their GenAI lessons.
Key Findings: The study revealed several important insights about GenAI integration in high school classrooms. First, both teachers designed lessons that blended rather than separated the "object-of-study" and "subject-specific" approaches, suggesting that teaching with or about GenAI is not a clear bifurcation but rather involves foregrounding different aspects of AI in various activities. June's lesson focused on exploring GenAI tools as research aids, encouraging students to evaluate their affordances and limitations while connecting to AP Research standards. Her students discovered cultural biases in AI image generation and raised ethical questions about data privacy organically during their explorations. Margot's lesson emphasized developing students' agency over GenAI writing tools by having them identify differences between human and AI-generated texts, critique AI writing quality, and provide feedback to ChatGPT. Her students concluded that while AI writing was more organized, it lacked the personal voice and specific evidence that characterizes high-quality human writing.
The co-design experience proved crucial for both teachers' lesson development. June significantly restructured her lesson based on cohort feedback, removing ethics discussions and frontloading exploration activities. Margot incorporated suggestions to have students compare human and AI writing side-by-side and added a ChatGPT feedback activity after observing other teachers' rehearsals. Both teachers reported that the professional learning activities helped them anticipate student reactions and develop confidence in teaching with unpredictable AI tools.
Implications: The findings contribute significantly to the field of AI in education by demonstrating that teachers do not need extensive technical AI knowledge to create meaningful GenAI learning experiences. Instead, they can leverage their existing pedagogical content knowledge and disciplinary expertise when supported by collaborative professional learning communities. The study suggests that effective AI integration requires teacher-centered approaches that prioritize pedagogical design capacity over technical training. The research also indicates that traditional frameworks for understanding curriculum design may need expansion to account for external design resources like communities of practice and AI agents themselves. Furthermore, the study shows that GenAI integration inherently involves elements of co-learning, where teachers and students explore AI capabilities together, requiring educators to embrace uncertainty and develop improvisational skills.
Limitations: The study acknowledges several important limitations. The small sample size of five teachers in the CRAFT cohort, while enabling intensive support and detailed analysis, limits the generalizability of findings. The researchers did not observe actual classroom implementations, relying instead on teacher reports and artifacts, which may not capture the full complexity of student-AI interactions. The online format of professional learning sessions limited teacher interactions to 60-90 minute moderated discussions, potentially constraining the depth of collaborative design. Additionally, the teachers were geographically dispersed across four time zones, preventing in-person observations that could have provided richer accounts of student learning and classroom dynamics.
Future Directions: The researchers recommend several areas for future investigation. Studies should extend data collection into actual classrooms to observe student-AI interactions and learning outcomes directly. Longitudinal research is needed to examine how teachers' confidence and GenAI integration practices evolve over time, especially as tools and institutional expectations continue to shift rapidly. Future work should explore the design of professional learning activities that simultaneously build teachers' understanding of AI and their subject-specific integration skills. Research should also investigate how different school contexts, policies, and resources influence GenAI integration approaches. The authors suggest examining how collaborative knowledge-building among teachers can be enhanced and sustained beyond formal professional learning programs.
Title and Authors: "Teaching high school students about generative AI: Cases of teacher lesson design" by Victoria Delaney, Ibrahim Oluwajoba Adisa, Christopher Mah, and Victor R. Lee.
Published on: June 11, 2025
Published by: The Journal of Educational Research