At the forefront of Duke’s AI education initiatives, Jon Reifschneider is driving innovation and impact. He is the executive director of the Master of Engineering in AI for Product Innovation program and co-founder and CEO of Inquisite AI, a Duke spin-out company. Through his teaching, research, and entrepreneurial efforts, Reifschneider is shaping the future of how AI can be thoughtfully and ethically integrated into learning environments.
Preparing Students for the Real World of AI
In his leadership of the AI Master of Engineering program, Reifschneider focuses on preparing students for the demands of industry. In his courses on large language models (LLMs) and machine learning, he thoughtfully balances where and how his students can use generative AI to assist their learning
In open-ended projects, he encourages students to use AI tools to accelerate their progress—particularly in coding-intensive assignments where the goal is iterative improvement rather than simply arriving at a correct answer. Built-in tools that assist with coding help students make quicker, more meaningful strides in their work. However, Reifschneider is careful to draw clear boundaries: in his Machine Learning course, for instance, generative AI is not always permitted, as it could provide solutions too easily and allow students to bypass critical thinking.
This nuanced approach models how instructors can harness AI’s benefits while ensuring students engage meaningfully with complex concepts and real-world scenarios. It equips them with critical skills they will need to navigate a professional world where AI is quickly becoming an essential tool.
Innovating with AI to Support Learning
Reifschneider is also advancing how AI can directly support education through applied research. At the Center for Research in Engineering in AI in Education (CREATE), he and his team have developed Qubit, a coding assistant designed with a Socratic approach. Rather than giving students direct answers, Qubit watches as students write code and provides hints and guidance based on their needs.
Qubit, currently available through Visual Studio Code’s extension marketplace for Duke students, allows users to adjust how much help they receive, making it particularly effective for beginning coders. A pilot program with high school students last summer showed that those who used Qubit had a more positive attitude toward coding and were less likely to give up when faced with challenges—a promising indicator for broader applications.
Teaching Students—and Faculty—About AI
Reifschneider notes that comfort with AI varies widely across society, with some—like his engineering and computer science students—already engaging with the technology confidently, while others are still developing an understanding of its capabilities and limitations. Therefore, he actively engages not only with students but also with broader audiences—such as community centers and retirement communities—demystifying how AI works and helping people appreciate both its power and its boundaries.
One key concept he emphasizes is iteration: the idea that working with AI is an evolving, back-and-forth process. He encourages learners to recognize that models like Anthropic’s Claude, which he personally favors for class preparation, can infer and anticipate much, but still requires human judgment and refinement.
Jon Reifschneider giving a presentation on AI to a local retirement community.
Advice for Instructors Beginning to Explore AI
For faculty considering how to navigate AI in their teaching, Reifschneider’s advice is clear: lean into it. “You can ignore it, but your students are using it,” he explains. To guide responsible and ethical use, he recommends setting clear policies outlining what is and isn’t allowed, integrating AI into assignments where it can help build valuable skills, and spending time getting personally comfortable with AI tools before bringing them into the classroom.
He also encourages instructors who are new to AI to start with bite-sized training sessions—just five to ten minutes at a time—to gradually build their confidence. If you are looking to get started, try Ethan Mollick’s short video series Practical AI for Instructors and Students, which introduces key concepts, demonstrates effective prompting, and offers examples of how AI can support both teaching and learning. Additional resources from Duke on teaching with AI, including LILE’s teaching guides, are available at the AI at Duke website.
Through his leadership, research, and candid conversations, Jon Reifschneider is helping the Duke community navigate the opportunities and challenges of a new AI-powered era—one thoughtful, practical step at a time.
