Lessons Learned from Developing New AI Courses for MBA and Undergraduate Business Students
Over the past year I designed and taught a new MBA elective, AI Methods for Social and Visual Data, and I was developing an undergraduate course, Artificial Intelligence for Business Leaders, launching that fall. In this talk I shared what has worked well in the MBA classroom, from assignment design to helping students build hands-on skills with modern AI tools, along with what I was changing or trying differently in the new undergraduate course. I also gave a brief introduction to a large randomized controlled trial of AI in the classroom that I am taking part in, and what we hope to learn from it about how AI actually affects student outcomes.
The session closed as a discussion with the faculty in the room: what they would add to the undergraduate course, what did not fit, and how Tepper should approach a flagship AI course for undergraduates.
Selected slides
From 45-884: MBA students learn how Vision Transformers apply attention to image patches.From 45-884: a joint embedding space where images and text sit side by side.From 45-884: the Tesla Vision versus Waymo case, cameras against lidar and radar.How 45-884 maps to Bloom's taxonomy: quizzes for remembering, labs for applying, the final project and AI in the News for creating.From 70-445: muffin or chihuahua? You can spot it. Now write the rule.From 70-445: agent = model + harness, the idea behind the course's agent unit.
June 5, 2025 · Teaching with AI Summer Workshop · Kellogg School of Management, Northwestern University
AI Data Visualization Coach
A flash talk with my colleague Zoey Jiang on the custom GPT we built for Data Visualization. Before presenting a chart redesign, teams test it with the coach. It will not hand over a redesign. It questions the team from three seats: a journalist, a chart designer, and a business stakeholder.
The experiment: treatment teams revise news charts with the AI coach before presenting; control teams revise without it.Why feedback can backfire: Kluger and DeNisi's model of where feedback helps and where it hurts.A chart a team brought in, with the strengths and problems they found.The same data after the coach's questions: a diverging bar, direct labels, and color that carries meaning.The coach's recommendation on another team's chart, with what improved and what still needs work.What students reported: 97 percent said the coach gave them incorrect or conflicting feedback at least once, which is part of the lesson.
April 30, 2026 · Tepper School of Business
Faculty Spotlight: what students learn in the MS in Business Analytics
A conversation for Tepper about the MS in Business Analytics curriculum, and why I describe business analytics as a decathlon.
