When Dr. Jessica Pryor redesigned her graduate educational technology course around Google Gemini Gems,customizable AI chatbots that deliver personalized, interactive content,she expected setbacks. Instead, students began building their own AI-powered tutoring bots, substitute teaching assistants, and interactive learning tools for their classrooms, signaling that the experiment was working.

The shift reveals a different approach to AI in higher education: rather than asking whether AI should teach, educators are asking how to teach students to use AI thoughtfully. Pryor's EDU 626 course, part of Murray State University's Instructional Technology Endorsement program, offers a case study in what happens when professors prioritize hands-on experimentation over abstract theory.

What Made This Course Different From Traditional AI Education?

Pryor's course began with ethics, not tools. Before students interacted with any AI chatbot, they studied the Family Educational Rights and Privacy Act of 1974 and discussed responsible AI practices. This foundation shaped how students approached the technology throughout the semester.

The course structure itself broke from convention. Instead of lecture videos and written notes uploaded to a learning management system, Pryor and Casey Stubblefield, a recent computer science graduate, transformed lesson plans into what Stubblefield called "a game-style narrative that students could actively move through." The interactive format meant students weren't passively reading information; they were exploring concepts as part of an engaging learning experience.

One critical design choice set this course apart from many AI education initiatives: Pryor kept human judgment at the center of assessment. AI chatbots delivered content and personalized learning support, but grading and feedback remained entirely human-driven. "Feedback is where the real learning happens," Pryor explained. "As a professor, I need to understand what my students are understanding. I don't want a bot to do that".

How Did Students Respond to Learning With AI?

The response surprised even Pryor. Students weren't just consuming course material; they were building. Although creating a chatbot was never a course requirement, students began developing their own AI-powered tools for practical classroom use.

The turning point came when a student reached out with a technical problem. "My first moment of 'wow, we did something really cool,' happened when a student reached out and said, 'I'm having trouble getting my bot to stop overexplaining the topic to my kids,'" Pryor recalled. "That's when it occurred to me that this was working,they were building their own bots for use in their classrooms".

Student projects included:

• Tutoring Chatbots: AI tools designed to help students learn specific subjects without overwhelming them with unnecessary detail
• Substitute Teaching Assistants: Bots configured to handle routine classroom questions and free up teacher time for deeper instruction
• Interactive Book Characters: AI-powered characters that students could interact with while reading literature or exploring narratives
• Simulated Patients for Health Science: AI systems that role-play medical scenarios for nursing and health science students to practice clinical reasoning

Why Does Teacher-Led Curriculum Design Matter for AI Literacy?

Across higher education, a parallel effort is underway to reshape how AI is taught. Stanford University's CRAFT initiative, developed through collaboration between Stanford researchers and practicing high school educators, takes a similar approach: treating teachers as curriculum co-designers rather than implementers.

The reasoning is straightforward. Schools are adopting AI tools faster than they are deciding what students should understand about them. Students learn to prompt systems for answers, but they don't automatically learn how models are trained, how outputs can mislead, or how data choices shape results. Those insights require instruction.

"Incentives in the private sector favor promoting their tools in educational programs, when there is a large and complex ecosystem of tools that will come and go, so keeping AI literacy focused on enduring and durable ideas and practices that are not bound to specific tools is important," said Victor R. Lee, Faculty Lead for the Stanford Accelerator for Learning's Initiative on AI and Education.

Victor R. Lee, Associate Professor, Stanford University Graduate School of Education

When curriculum is built outside the classroom by tool vendors or external consultants, teachers become implementers rather than authors. CRAFT intervenes at that structural layer by involving educators from the beginning. Their experience managing time constraints, varied learning goals, and diverse student backgrounds informs the design from the start.

How Can Educators Build AI Literacy Into Existing Courses?

For teachers looking to integrate AI education without overhauling their entire curriculum, several practical strategies emerge from these initiatives:

• Start With Ethics: Before introducing any AI tool, establish a foundation in responsible use, data privacy, and the limitations of automated systems. This frames AI as a tool with tradeoffs, not a magic solution
• Design for Adaptability: Create lesson materials in editable formats like Google Docs rather than static PDFs. Teachers need the ability to adjust terminology, add examples, or remove sections to match their class level and pace
• Connect to Existing Learning Goals: Rather than treating AI as a standalone unit, integrate it into writing instruction, scientific reasoning, media literacy, or other subjects already in the curriculum. This prevents AI from displacing existing objectives
• Prioritize Hands-On Experimentation: Students learn more from building and troubleshooting their own AI tools than from passive exposure. Create space for students to experiment, fail, and iterate
• Keep Human Judgment Central: Use AI to deliver content and provide initial feedback, but reserve grading, assessment, and deeper feedback for human educators who understand their students' learning needs

Lee emphasized that effective AI education depends on interpretation as much as availability. "There is a lot to know about whether and how to use the tools effectively and to ensure that the value for what teachers and students care about is there," he noted. Teachers need ways to connect technical behavior to classroom learning goals, not just access to platforms.

Lee

The broader challenge is structural. Schools operate on tight schedules, and instructional time is already committed to required subjects, grading, and administrative work. Asking teachers to redesign curriculum requires institutional support, professional development time, and recognition that this work is part of their core responsibility.

Yet the payoff appears significant. When educators help build lessons, AI concepts can align with existing curriculum without displacing other objectives. Students gain durable knowledge when instruction explains how systems work rather than simply introducing platforms. The central question shifts from whether students can access a tool to whether they can evaluate and interpret the tool's output.

For Pryor, the project ultimately became less about artificial intelligence itself and more about innovation, curiosity, and reimagining what teaching and learning can look like in higher education. "Innovation does not always start with expertise," she said. "It begins with curiosity".