‘We Want To Define The Future’: Inside CMU Tepper’s AI-Powered Transformation

At Carnegie Mellon University’s Tepper School of Business, artificial intelligence isn’t just a topic of conversation — it’s the new language of learning.

Since early 2024, Tepper has rolled out an ambitious AI program that goes well beyond the use of ChatGPT in the classroom. Under the school’s Collaborative AI Initiative, faculty and staff have developed 10 custom-built, AI-powered “Interactive Learning Labs” that simulate the real-world ambiguity, stakeholder dynamics, and judgment-heavy decisions students will face on the job.

“These labs aren’t just about using AI tools,” says Param Vir Singh, associate dean for research and the Carnegie Bosch Professor of Business Technologies & Marketing. “They’re about students learning how to think, act, and adapt in a world where AI is part of the team — and part of the problem-solving process.”

THREE ROLES FOR AI: COACH, STAKEHOLDER, AND SIMULATOR

Each of Tepper’s AI labs uses generative AI in a highly structured, scaffolded way to deliver learning outcomes while preserving pedagogical integrity. Singh outlines three distinct AI roles embedded in the labs: Socratic coach, stakeholder persona, and environmental simulator.

“The Socratic coach challenges students with questions that lead to reflection and deeper learning — but never gives the answer outright,” Singh tells Poets&Quants. “Meanwhile, AI personas stand in for stakeholders like CEOs or investors with distinct agendas. And simulators create dynamic environments — competitors, customers, or market shifts — so students must make and iterate on decisions in context.”

Taken together, the system replicates the complexity of real-world decision-making — but in a safe, adaptive, and data-rich format that gives both students and professors powerful feedback loops. 

“Faculty can see everything students are doing,” Singh says. “Every prompt, every keyword, every strategic choice — it’s all trackable and usable in post-case debriefs.”

FROM HEADPHONES TO HEDGE FUNDS: WHAT THE LABS TEACH

The AI labs span multiple disciplines: marketing, finance, operations, communication, and business ethics. In a Search Ads Strategy Lab, for instance, students design and launch a Google Ads-style campaign for noise-canceling headphones. They interact with an AI market research agent trained on realistic consumer data, extract insights through smart questioning, and then build and optimize keyword and ad strategies.

In another lab focused on Portfolio Management, students meet with an AI-modeled investor, deduce their risk preferences through dialogue, and construct tailored investment portfolios using live Yahoo Finance data. “This is very different from just being told the risk tolerance is X,” Singh says. “In the real world, you have to figure that out through conversation.”

Other labs include a Feedback Coach for improving peer evaluations in communication classes, a Design of Experiments simulation where students troubleshoot production flaws in a paint factory, and an ethics module where students engage with each other and an AI mediator on morally sensitive business scenarios. Still in development: digital twins of venture capitalists, allowing students to pitch startup ideas to AI-trained replicas of real investors.

LESS LECTURE, MORE JUDGMENT

The push toward AI-integrated learning at Tepper is grounded in two major disruptions Singh sees reshaping graduate business education.

“First, we know that a significant portion of graduates’ future jobs — or at least job tasks — will be automated,” he says. “Second, entry-level roles that once served as apprenticeships are disappearing. Employers expect students to be client-ready from day one, with better judgment and higher-level skills.”

That creates a pedagogical challenge — and an opportunity. “We can’t wait for industry to tell us what to teach,” Singh argues. “Business schools have traditionally followed industry. We want to define the future alongside it.”

THE INFRASTRUCTURE BEHIND THE INITIATIVE

Building these AI-driven learning experiences wasn’t as simple as plugging in off-the-shelf tools. Singh, who describes himself as “a very techie person,” quickly realized that customized, domain-specific engineering was required. Tepper hired a full-time AI engineer and began working closely with faculty to match the lab architecture to learning objectives — with safeguards in place.

“You can’t just create a custom GPT and expect it to teach your case,” he says. “You need scaffolding to avoid gaming, prevent hallucinations, and make the experience repeatable without being robotic. Every agent is trained with our own case data to deliver rigor and consistency.”

That level of sophistication has made Tepper a point of reference. At the recent IUGBE 2025 Conference at Rutgers Business School — themed “The Futureproof Business School” — Singh says deans and faculty from other institutions were “hungry” for insight. “Most schools are still at the stage of giving faculty incentives to experiment with AI tools,” he notes. “We’re already building full simulations — with measurable impact.”

STUDENT IMPACT: ‘VALUE FOR THE TUITION DOLLAR’

Perhaps most importantly, Singh says the AI labs are paying off in the classroom.

“When I ran the Search Ads lab in my course, the students loved it. There was a huge range in performance — but everyone was learning, and the faculty could pinpoint exactly where the gaps were,” he says. “That kind of insight is rare. This is real value for the tuition dollar.”

The model, he believes, is scalable — but only with intention and support. Tepper isn’t just dropping AI into syllabi. It’s embedding it in pedagogy, coaching, assessment, and student growth.

“We’ve built ten of these already in just over a year,” Singh says. “Because we didn’t want to wait. Industry is uncertain right now. But we see that as a chance to lead — not follow.”

DON’T MISS TEPPER SCHOOL DEAN STEPPING DOWN NEXT YEAR

© Copyright 2025 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.