Generative AI and Pedagogical Innovation: Two Students’ Perspectives

CARADITE is Duke’s think-tank and living laboratory for innovative research and design about transformative education. Earlier this fall, our Center’s undergraduate research assistants–Emma Ren and Barron Brothers–participated in the second annual Emerging Pedagogies Summit. Here, Emma and Barron share their thoughts about how the Summit addressed the role of AI in education, with an emphasis on AI as “a tool to enhance—not replace—human relationships.” Enjoy! – Remi Kalir, PhD, Associate Director, CARADITE

The Emerging Pedagogies Summit at Duke University brought together educators and researchers to explore how generative AI and innovative educational strategies are transforming teaching and learning. As the only two undergraduates in the room, we found the discussions insightful and practical for bridging the gap between traditional educational practices and AI-driven learning tools. The Summit challenged the common assumptions about AI in education, offering new perspectives on integrating these technologies to enhance the student experience while prioritizing critical thinking and human connection at the center.

On the Summit’s first afternoon, we participated in Carter Zenke’s workshop exploring the potential of generative AI, specifically the use of GPT (Generative Pre-trained Transformer) as an educational tool rather than a threat to learning. GPTs are models trained on massive datasets to generate human-like responses based on input prompts, extending beyond just ChatGPT. AI tools like these can brainstorm, summarize, and generate responses that mimic human conversation.

Carter Zenke listens as an audience member asks a question.

In his previous work at Harvard, Zenke showcased CS50.ai, a pedagogical AI-based teaching assistant that guides students through coding problems step-by-step. Instead of providing a direct answer, this AI tool prompts students with questions, fostering a learning environment akin to “rubber duck debugging,” a problem-solving technique whereby articulating problems aloud can reveal a solution. This tool highlights how AI can assist students during challenging moments, encouraging critical thinking while receiving patient and personalized feedback.

We also learned about how techniques like Retrieval-Augmented Generation (RAG) can expand AI’s utility by customizing responses using uploaded course material. This approach can reduce the risk of misinformation and bias, better providing students with access to accurate, course-specific information. These capabilities can make AI a powerful resource for personalized learning, offering the opportunity for both synthesis and brainstorming.

When, for example, an instructor wants to direct how an AI responds to a student (i.e. ensuring it doesn’t solve the problem for a student outright), related system prompting approaches can help. Unlike user prompting, system prompting affects how GPT can respond to a user prompt. System prompting is helpful in reducing the number of answers given by GPT and in generating a nuanced, student-led discussion. We found that telling a GPT to “encourage the student to respond”, “avoid breaking down the steps… right away”, and “focus on making the student UNDERSTAND the abstract” could transform how this technology encourages active, nuanced learning. System prompting can support students to receive helpful guidance rather than simple answers, which can move the discussion around AI from fear to informed implementation.

We’ve seen many course syllabi that ban AI outright, equating its use with plagiarism. AI can do much more than simply giving us students the answer. The Summit encouraged innovative AI usage in the classroom rather than treating it as taboo. While some educators might fear that AI tools undermine our learning by providing instant answers, the Summit encouraged a shift toward AI use for teaching and learning support.

In his workshop, Zenke further emphasized the value of AI as a patient, judgment-free partner for students. The “inhuman level of patience” that AI offers can be particularly beneficial for introverted students who may feel self-conscious asking questions in front of their class. AI may especially benefit students who prefer to work independently or need more time to process material. That said, the Summit also stressed the importance of balancing AI use with meaningful human interactions. While AI can handle certain repetitive tasks, personal connections between students and educators remain essential for fostering skills like empathy and collaboration. Essentially, AI is a tool to enhance—not replace—human relationships.

The Summit was a valuable reminder that innovations in both AI and education need not exist in opposition. With thoughtful integration by educators, AI can offer personalized support to students, enhance accessibility, and foster deeper learning. However, the Summit also emphasized the importance of maintaining a balance whereby AI is a tool used to complete routine tasks. As students, we believe it’s critical that our instructors preserve space for personal interaction, exploration, and critical thinking.

Emma’s Final Thought: Truthfully, I believe the impact of GPTs extends beyond merely being a tool, and if misused, it could cause significant harm. However, since GPTs are here to stay, whether educators welcome them or not, they should focus on leveraging them effectively. If we choose to ban GPTs entirely instead of embracing them, we risk putting students and teachers on opposing sides; students may resort to using GPT tools in secret, which don’t always provide the most accurate, credible, or helpful responses, and teachers will struggle to prevent AI use altogether. Rather than creating this divide, educators should aim to harness GPTs’ potential as “personalized, patient-guiding mentors” to enhance learning and academic collaboration.

Barron’s Final Thought: When we students are in a structured classroom environment, it’s sometimes difficult to gain our instructors’ perspectives on these emerging pedagogies. Not only was the Summit great at bringing us and many faculty together, but it also provided the opportunity to talk with these faculty about the future of our education, especially with AI. I am glad that these discussions were both sensible and Janus-faced—instead of coming from a place of fear, we discussed how AI could be used practically as a tool to improve learning. I was also glad that I had a voice in the Summit; I do not have the experience of a professor, but my thoughts were still heard and validated. In this way, I felt like I played a bigger role in helping Duke be the best it can be.


About the Authors

Emma Ren is a sophomore interested in philosophy (her major) and music. At CARADITE, she is excited to explore the ecological impact of AI in reshaping cultural norms and academic practices, especially by diluting credibility in writing and eroding engagement with complex ideas. 

Barron Brothers is a junior studying statistical science and music. At CARADITE, he plans to explore the validity and ethics of generative AI, specifically how AI-assisted study techniques enhance and preserve the student learning experience.