At Fuqua, 3 Professors Rewire The MBA For The AI Era

Three professors at Duke University’s Fuqua School of Business are remaking what an MBA classroom looks like, each from a different angle. 

One is pulling AI out of the technology stack and into the realm of society, ecology, and strategy. Another is pushing students to interrogate where AI helps and where it quietly fails them. The third is using AI to recover something business schools have been losing for years: the texture of real human interaction.

Dan Vermeer, associate professor of the practice and executive director of Fuqua’s Center for Energy, Development, and the Global Environment, spent 18 years building the school’s platform on climate, energy, and sustainability. This year he launched a new mini-course on AI that he wrote, in part, while teaching it. “It’s a mini-course, so it’s six sessions,” he tells Poets&Quants. “It was a big topic to take on in successions, kind of a teaser class.” Roughly 75 students signed up the first time it ran, many of whom had never appeared in his sustainability classes. Attendance held at ninety to ninety-five percent across the term.

Vermeer’s approach is deliberately not technical. “The class is very much not trying to teach them how to use AI. They should learn how to use AI. There’s places for them to do that.” His focus is what he calls externalities, the social, environmental, and strategic shocks AI is creating for business leaders. “I want to be able to build their capacity to respond creatively even though they can’t predict what those are going to be.”

CASES REWRITTEN ON DEADLINE

Vermeer wrote his own case studies and revised them on the fly.  One on AI governance, built around Anthropic, was rewritten three times because the news kept moving. He carved out the first 10 minutes of every class for what he called a news flash segment, in which two students presented a current article using a template that asked, among other things, whether business leaders should have seen the development coming. “It was a way to fill in the gaps of the landscape that they care about that we weren’t covering in the class,” he says.

Industry professionals have started showing up, too, drawn by questions their companies cannot answer alone. They are weighing high-stakes investments without clear returns, sorting out what to build versus buy, and watching competitors cut significant portions of their workforces. “They’re confronted with a whole set of challenges around things like workforce, like what happens when my competitors are laying off huge chunks of their workforce. I think they’re worried about things like what happens if my brand gets taken down in a deep fake attack overnight.” Productivity data, he notes, is still mixed.

Vermeer is candid about the broader challenge facing business schools. Fuqua, he says, has embraced AI as a tool for transforming the learning experience, but the deeper integration of AI into society is still under-covered. “It’s not exactly the ethical questions. It’s more the strategic questions. And that’s kind of what I tried to highlight in my class.” His grounding in climate, energy, and water studies gives him an entry point that purely technical courses lack. He has built what he calls the AI planetary stack, a framework that places the technology at the surface and stacks the organizational, economic, geopolitical, and planetary layers beneath it. “That’s going to be the same looking out decades,” he says. “That doesn’t change. The stuff that flows across and where the connection points are, that’ll change, but the landscape, I think, is pretty stable.”

Students push back, and he wants them to. The biggest debate of the term, he says, was over AI and military applications, prompted in part by Anthropic’s position on the issue. Workforce implications and investment levels surfaced repeatedly. “It was actually part of the design was to make that happen,” he says.

A DIFFERENT ROUTE IN

Elia Ferracuti, associate professor of business administration in Fuqua’s Accounting area, has taken a different route into the same set of questions. He has rebuilt his managerial accounting course around three uses of AI. 

The first is knowledge acquisition. A virtual teaching assistant, trained on his own course materials, will answer questions, generate practice problems, and toss in the occasional Italian word so students find it approachable rather than dry. A set of widgets lets students manipulate core concepts like break-even analysis. A “time travel” bot, built around an old chemical-bags case from the 1980s, lets students ask about the differences between that business environment and today’s.

The second pillar is simulation. Ferracuti no longer hands students the financial data they need to make decisions. Instead, they have to extract it from a bot called The Accountant, modeled on the no-nonsense character from the film of the same name. 

“If a student says, I want data to make a new product introduction decision, the answer would be, ‘That question is too generic. Don’t waste my time,’” Ferracuti says. “What data do you want?” The point, he says, is that knowing what to ask is itself a managerial skill. Next year he plans to build a transfer pricing exercise in which a student takes on different roles in a multi-party negotiation, with two bots playing the other roles. 

“In that sense, you get exposure to the different possibilities. Maybe you are HQ, maybe you are a service provider, maybe you are a service receiver.”

HUMAN FIRST, AI FIRST

The third pillar is the most experimental. Across 10 case studies, Ferracuti rotated three approaches: human only, human plus AI as a support tool, and AI first with human revision. 

The results surprised him. 

Solutions were reasonable under every approach. But the human-first submissions made their assumptions explicit and offered a wider variety of answers. The AI-first submissions did more sensitivity analysis and applied more managerial accounting tools, but they were more homogeneous and less explicit about their reasoning. 

“I told them, if you are a consultant, that’s something you want to think about,” he says. “You go pitch to a company, if you propose the same thing as others, maybe you’re not going to win.”

The biggest surprise was the failure mode. “The average quality with AI was good,” Ferracuti says. “But when it was wrong, it was massively wrong.” He points to a case where an overhead allocation rate was rising over time. The AI told students to invest in new internal information systems to fix the measurement. In fact, the rising rate was a symptom of a deeper problem. “What AI did was misinterpret a high fever for having an infection. So the solution was, ‘We need to lower the fever,’” he says, “not, ‘We need to remove the infection.'” Many student teams proposed treatments for the fever and never touched the underlying issue.

The misconception Ferracuti hears most often has less to do with AI than with what managerial accounting actually is. “They imagine you come into a managerial class, you apply a formula that is given to you, the formula producing numbers,” he says. “So as a consequence, I can use chatbots and I can say, this is my problem. Apply the formula and give me the solution.” In practice, he said, most managerial decisions have many acceptable answers and a few catastrophic ones. “Without the thinking, it’s not going to end up in a good place.”

BRINGING THE TEAM BACK

Scott Dyreng, senior associate dean of innovation and professor of accounting, is using AI to address a problem that predates ChatGPT. Team meetings, he says, have been deteriorating for years. 

“Teams, over time, many years, have been meeting together less and less, and dividing and conquering more and more,” he says. When students were given unrestricted access to AI, the meetings got worse. “Why do I need to go meet with my team? I just meet with my AI chatbot.” 

Over Christmas break, Dyreng started designing tools that would push the other direction. He laid out the case for it in a short Wall Street Journal piece in February, framing AI as a way to bring teams back, not to dismantle them.

The first piece is participation tracking in the classroom. Faculty have long struggled to score participation fairly, Dyreng says. Some hire teaching assistants, some try to remember at the end of the day, some wait a week and reconstruct it. The AI-enabled classroom, which Dyreng and colleagues developed and rolled out between mid-January and late February, now captures ninety-five to one hundred percent of student comments and creates an audit trail. “If a student says, well, I don’t remember that, we could say, well, here’s what the transcript said,” he says.

The second piece – and the one that excites him most – is team-meeting feedback. A student places a phone on the desk, hits record at the start of a meeting and stop at the end. The system then identifies who said what and generates an analysis: who took airtime, who made substantive versus procedural comments, who built on others’ contributions, whose voice was pushed aside. Each member receives the feedback, and the team can adjust before the next meeting. After six or seven meetings, the system can deliver individualized analysis: tendencies, strengths, weaknesses, and suggestions for improvement.

Faculty get a parallel report. Dyreng can walk into class knowing which concepts students wrestled with in their team prep, who in the room understands a particular idea well enough to teach it, and which teams reached divergent conclusions worth surfacing in discussion. He can also analyze his own teaching after the fact. “I got feedback from a student that suggests that I do X. Can you analyze every interaction I have and tell me, do I do X and how common is it and when do I do it?”

‘I CAN SHOW YOU WITH DATA’

Dyreng is careful about the data question. Fuqua has video-recorded every class for years, he says. “This essentially is not gathering any more data. We’re using the same data. We’re just using AI to make it useful.” Names are stripped before transcripts are sent to AI models for analysis, and results are reassembled locally before they reach students.

The longer-term promise, he says, is what students will be able to do with the accumulated record. A student starting Fuqua in the 2026 to 2027 academic year will, by graduation, have a time series of feedback across accounting, marketing, strategy, operations, and economics. 

“When I go to my interview, somebody says, ‘What are your strengths and weaknesses?’ I can say, I know exactly what they are. I’m really good at leading a team discussion if it’s in marketing. I’m catching up in finance. I’ve been taking these courses to help me be better,” Dyreng says. “I can show you with data.”

There is also a subtler benefit, Dyreng says, in the fact that AI is the messenger. When a professor tells a student their participation needs work, “they think of me as the enemy now.” When AI delivers the same feedback five times in a row, the student often comes to the professor and asks for help. “Suddenly I’m their advocate instead of their adversary,” he says.

‘WHAT WE CARE ABOUT IS THE HUMANITY OF WHAT WE DO’

Across the three classrooms, the throughline at Fuqua is the same. AI is forcing sharper questions about strategy, judgment, and what only humans can do. 

Vermeer is building frameworks designed to hold steady while the technology churns. Ferracuti is teaching students that the average AI answer is fine, and that the fine answers can also be the wrong ones. Dyreng is using AI to rebuild the social architecture of the MBA program rather than erode it. 

“What we care about here is the humanity of what we do,” Dyreng says. “That will never change. And then what we want to do is figure out how that interacts with whatever technology is available.”

DON’T MISS MEET DUKE FUQUA’S MBA CLASS OF 2027 and HOW DUKE FUQUA IS USING AI TO TRANSFORM THE BUSINESS CLASSROOM

© Copyright 2026 Poets & Quants. All rights reserved. This article may not be republished, rewritten or otherwise distributed without written permission. To reprint or license this article or any content from Poets & Quants, please submit your request HERE.