A Business School Professor’s Lessons From Teaching AI To MBAs

Last semester, I taught a Professional MBA course on artificial intelligence in marketing. By the end, my students and I had both been transformed by it.

It was the most intellectually intense semester of my career – more urgent, in some ways, than my work at PricewaterhouseCoopers or Deloitte – and I truly enjoyed it. But teaching a subject that changes faster than anyone can fully keep up with also made something clear: AI education is a burning present challenge, not a future one.

First, we need to stop categorizing AI as “just a tool.” A tool is something you use, but it doesn’t decide or act. The systems my students worked with did not behave that way. One student gave an AI agent a plain-English instruction and watched it generate leads and draft personalized outreach emails. Another built an automated competitor-monitoring system that tracked rival websites, compiled weekly intelligence reports, and delivered them by email each Monday. These were not people simply using tools. Rather, they were people working together and supervising a new kind of intelligence.

Realizing this distinction changes what we need to teach. The main skill to teach is no longer prompt engineering. It is judgment in knowing when to delegate tasks, when and how to intervene in the process, how to verify the output, and when to override a system that sounds right but may be wrong. If we treat AI as a tool rather than an intelligence that requires management, we will fail to develop the judgment needed to tell the difference.

The exceptionally fast pace of change makes the problem harder. Some of the materials I prepared on Tuesday would be outdated by Thursday. In a matter of weeks, new capabilities appeared such as connectors that let AI access external applications, reusable skills for complex tasks and workflows, and coding environments that could build functional business applications in minutes. Each of these updates could have filled one class session. If I taught only what I planned in December, students would have learned a version of AI that no longer existed by February.

That volatility extends beyond the classroom. Retailers are already experimenting with letting customers buy products through chatbots. Some platforms are developing ways for AI agents to transact on behalf of merchants. For professionals who manage teams, run companies, and allocate budgets, these developments are directly relevant. They threaten the assumptions their current strategies are built on.

Teaching this material also carries an emotional burden. Every week, I showed students how a system could complete a task in minutes that might once have taken a junior marketing associate weeks to finish. I had the consistent tension of saying, “This is exciting,” and acknowledging, “This may cost someone their job.” The ethical questions are also hard to resolve. In a final-class demonstration, I generated music with a singing voice that closely resembled a famous artist’s. I could not be sure whose creative labor that output was benefiting from, with no credit given to the original creator. These systems are not neutral instruments; they are sociocultural artifacts shaped by the data, assumptions, and disputes embedded in their training.

Universities and companies treat AI adoption as a checklist: learn prompting, add automation, build workflows, deploy agents, and you are done. The focus should be on how we frame problems with AI. For any task, the real question is which combination of model, app, and harness setup will actually get you the outcome you need and get you there safely. That is a strategic, creative, and essentially human question.

What gave me hope was watching students develop a mental model or instinct for answering it. They did not turn out to be experts in any single AI platform. Instead, they stopped treating AI outputs as answers and started treating them as drafts. They learned to check the work. They moved from being users to becoming collaborators, managers, and critical thinkers.

Educators should not wait for the field to settle because it will not settle, at least not anytime soon. Business leaders should stop treating AI as a software purchase and start redesigning workflows and transforming organizational tasks. And policymakers should recognize that access to capable systems and the knowledge to use them well are turning into a question of fairness.

I will teach this course again next year, and much of my current material will already be obsolete by then. But the core lesson will stick. I will meet this moment by teaching judgment instead of deference, redesigning workflows rather than applying quick patches, and paying closer attention to what AI now asks from all of us. Class is not dismissed yet.

Ayse Ozturk is an assistant professor of marketing at the Darla Moore School of Business at the University of South Carolina. She previously worked at PwC, Deloitte, and Peugeot, and teaches graduate AI courses and leads AI workshops. Her teaching has been covered in The Wall Street Journal and by OpenAI Academy.

© 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.