News

From AI Experimentation to Classroom Impact: Wharton Students Help Educators Put AI to Work

A student from the Wharton AI for Society Research Lab presents to a group of teachers.
Anjali Dhupam, W'27, C'27, presents at West Deptford High School

For educators navigating the rapid rise of generative AI, the challenge is no longer simply understanding what the technology can do, it’s deciding where AI can genuinely improve teaching and learning, and where human expertise remains essential.

That question is central to the Wharton AI for Society Research Lab (WASRL), a six-month interdisciplinary research program run by the Wharton AI & Analytics Initiative (WAIAI). The Lab brings together undergraduate and graduate students from across the University of Pennsylvania to develop and test AI resources for K–12 educators.

The Lab’s initial focus is professional development for high school teachers. Its projects include AI-supported learning modules, lesson-planning and classroom-design assistants, and other resources shaped by research into teachers’ needs, routines, and experiences with AI.

Recently, that work moved into a classroom through a collaboration with educators at West Deptford High School in New Jersey.

From AI Tools to Teaching Tools

At West Deptford, WASRL students introduced faculty to practical applications of generative AI, including Gemini, Gems, and NotebookLM. The session focused less on showcasing new technology and more on helping teachers decide how these tools might fit into their work.

The team began with an overview of AI fundamentals and governance, covering bias, hallucinations, verification, and student privacy. Teachers then watched demonstrations and worked with AI using materials from their own classrooms.

Esha Desai, WG’27, L’27, a Penn Carey Law and Wharton JD/MBA candidate who led the West Deptford collaboration, said the goal was to make AI more immediately useful for the teachers.

“Teachers came in with different levels of AI experience, and by the end everyone had a working quiz they had built themselves and could actually use in class,” Desai said. “Watching that shift from ‘AI is abstract’ to ‘I built this in 20 minutes’ was the whole point of the workshop.”

Mathematics teacher Jaclyn Falcone said the workshop expanded her view of what AI could do. Before attending, she had used Gemini and ChatGPT mainly to search for materials or gather information.

“Seeing it less as just a chat and more as tools” was one of her biggest takeaways, Falcone said. She saw how AI could help adapt existing materials into resources she could use with her students.

That possibility is particularly relevant for teachers whose time is already taken up by lesson planning and instruction. Falcone said having dedicated time to experiment helped her find uses she might not otherwise have considered.

Chris Bennett, Senior Fellow with WAIAI, and Brandon Krakowsky, Director, Data Science and Research, Wharton AI & Analytics Initiative, provide instruction
Brandon Krakowsky, Director, Data Science and Research, Wharton AI & Analytics Initiative, provides instructions with WAIAI Senior Fellow Chris Bennet.

Designing Around Teachers

Getting to that point required the WASRL team to learn from teachers before designing resources for them.

“The surprise for me has been how much teacher expertise the process demands,” Desai said. “We can’t build good AI-supported lessons without deeply understanding what teachers already do: their workflows, their instincts about students, their concerns about integrity.”

Claire Liu, WG’27, a graduate researcher with the Lab, said the team encountered that lesson during its early research.

“We went in with hypotheses about what teachers would want,” Liu said. A focus group revealed concerns and obstacles the researchers had not anticipated. In response, they redesigned a later session to give teachers more time to work with their own classroom materials.

That process reflects WASRL’s broader approach: research teacher needs, build prototypes, test them with educators, and revise them based on what teachers say and do. For Xuan “Leah” Liu, GEd’27, WASRL research assistant, the West Deptford workshop showed the value of that approach.

During the session, she introduced teachers to NotebookLM and demonstrated how they could use it with their own materials. At first, she was unsure how receptive the group would be. “Once we started, though, I was surprised by how actively everyone explored the tool,” she said. “There was a real sense of curiosity in the room.”

Finding the Right Role for AI

That curiosity does not mean treating AI as a solution to every educational challenge. A consistent theme emerged from both the student researchers and the West Deptford educators: AI is most useful when it supports human judgment rather than replacing it.

For teachers, that might mean spending less time adapting materials, drafting communications, creating rubrics, or developing lesson resources. For students, the question is more complicated.

“Learning often requires productive struggle,” Claire Liu said. “If AI removes the thinking an assignment was designed to build, making it ‘easier’ can undermine the entire point.”

West Deptford Assistant Principal Kelly Saia also saw value in tools that guide students through problems instead of simply giving them answers. She described AI as another resource teachers can use for differentiated instruction while remaining central to the learning process.

Falcone emphasized that effective AI use still depends on teacher expertise. Teachers need to recognize inaccurate outputs, and students need enough background knowledge to ask useful questions.

“It’s not really giving them everything they need; it’s giving them guidance,” she said.

For Leah Liu, the goal is not to introduce AI simply because it is available.

“The more important question is where AI can solve a real problem or remove a real barrier,” she said. “Ultimately, I’m interested in how people and AI can evolve together, with AI supporting human judgment and learning rather than replacing them.”

At West Deptford, that work began with a straightforward goal: giving teachers the knowledge, tools, and time to decide for themselves where AI can improve their work and their students’ learning.

This content was created with the assistance of generative AI. All AI-generated materials are reviewed and edited by the Wharton AI & Analytics Initiative to ensure accuracy, clarity, and alignment with our standards.