Michael Olenick: Learning Requires Friction

This is the latest in a series about how AI is changing business school education, short interviews with faculty at various B-schools focused on the use of AI in business and business education. See The Deck Is Dead and We Teach MBAs To Apologize About The Best Tool They Have.

Stefano Puntoni is the Sebastian S. Kresge professor of marketing at The Wharton School at the University of Pennsylvania, where he co-directs AI at Wharton, and co-author of Decision-Driven Analytics, a Wharton School Press book on making decisions with data. A behavioral scientist, Stefano studies the psychology of artificial intelligence.

His students are anxious. They watch the headlines about entry-level jobs in consulting and banking, exactly the jobs MBAs chase, and wonder what’s left for them. Puntoni’s answer is to flip the frame: too much of the AI conversation is about cutting costs, and cost only goes to zero. Growth has no ceiling. For a young person fluent in these tools, he argues, this is an amazing time to enter the job market, not a terrible one.

Below, we talk about all of that, plus a classroom chatbot built from lecture recordings that turns “I took some notes in class” into a tutor that was in the room, and whether Wharton’s new AI major goes the way of the “international business” majors of the 90s.

One issue is how much thinking is being outsourced to GenAI. Two of Puntoni’s Wharton colleagues, Gideon Nave and Steven Shaw, recently coined a term in a viral working paper: cognitive surrender, the tendency to adopt whatever an AI says with minimal scrutiny, overriding both intuition and deliberation. As Stefano puts it, learning requires friction, and a tool that removes all friction removes the learning with it.

One thought of my own: the tutor probably couldn’t exist in Europe. Recording every lecture and piping it into an AI feels like the kind of thing GDPR was written to prevent. I spent years in France and while the privacy rules there are well intentioned, they sometimes come at a cost, in this case keeping students from better teaching.

Q: What do you think business school actually teaches, at the core of it?

Mostly mental models. A student told me once that she went into a job interview, got handed a problem she had no idea how to solve, remembered a framework from class, and used it to work through the problem live. That’s what business education is. We’re not teaching science in the sense of fundamental truths. We’re teaching tools you can reach for under pressure.

Q: Your students are entering a job market everyone says AI is about to gut. What do you tell them?

They’re worried, and I understand why. Every conversation about entry-level jobs in consulting and investment banking, the things MBA students want to do, sounds ominous. But almost all of that anxiety comes from the cost side of AI, companies shaving half an hour here, removing a person there. I tell them to look at the growth side instead. For a young person with energy and real expertise in these tools, this is an amazing time to enter the job market. There’s a strange paradox underneath it: technical jobs are becoming less technical, non-technical jobs are becoming more technical.

Q: What is “cognitive surrender,” and why does it worry you?

It’s what happens when AI removes so much friction from thinking that you stop doing the thinking. You adopt whatever it says. My rule for students, and for myself, is that AI never writes, it only edits. I promise myself I’ll come to AI with ideas, that I’ve thought about something on my own first. Then AI can sharpen the idea or poke holes in it and show me my idea was bad. This year I added a course session just on this, on being deliberate about what you allow yourself to do with AI. The goal is twofold: build the skills you need to succeed in an AI-driven world and prevent the loss of skills from using AI.

Q: Do you think AI thinks the way we do?

We now have a second source of intelligence that is very different from ours, and we don’t fully understand the similarities and differences, partly because we don’t fully understand our own intelligence. Maybe the two stay fundamentally different forever. Maybe the things we consider uniquely human turn out to be replicable. It’s very difficult to be confident either way. Very smart people, much better informed than me, swear by one view or the other. It’s become a religious argument.

Q: That cuts against the usual pitch that AI makes everything easier.

Making things too smooth can be a problem in education. In learning, if AI becomes something that removes all friction, then probably there will be no learning. Learning requires friction.

Q: Where does AI hit business hardest?

Marketing, more than any other function. Most of the AI conversation in business is about productivity: more output per employee, cheaper, faster. That’s a cost story, and cost stories have a ceiling, you can only cut to zero. Marketing’s job should be to make the growth argument: new products, new customer relationships, new customer experiences, better margins, better brands. The marketing function should be the champion in the organization of seeing AI as a tool for growth.

Q: How is this changing what happens in your classroom?

Two things: content and delivery. Content changes every year, this year I added a session on agents. Delivery is messier. Some of it’s a loss: the take-home essay is dead because you can’t tell who wrote it, so a lot of professors are bringing back the blue book just to watch students write a page. But some of it opens things up. Last year I had every student build a working chatbot. The brief was as broad as I could make it: solve a problem of your choosing, in a manner of your choosing.

Q: Tell me about the Wharton classroom bot.

Every classroom computer starts recording when you log in, a COVID legacy. So we have video of every lecture. We extract the transcripts and feed them into a bot along with the readings, slides, and syllabus, inside the course’s ChatGPT environment. Students ask it anything about the course and it answers from the materials. But it also connects [to the platform] where the videos live, so when it explains a concept it hands you the clip where I talk about it, and you click and hear me for three minutes. Suddenly students don’t have notes, they have a companion: a tutor that can quiz them, explain things in language they understand, and find the exact moment in class where we covered it. You could do this teaching ancient history.

Q: How is AI changing marketing as a field?

Start with research. These models were trained on essentially everything, so they encode enormous knowledge about how people think and decide. That’s driven an explosion of interest in synthetic data and digital twins. We published an HBR piece on AI moderation, where bots run the interviews and code the qualitative data at a scale a human team never could. If you’re teaching marketing research today, synthetic data, digital twins, AI moderation, and coding unstructured data are now core to the field. And that’s before the go-to-market side: ad executions, social media, customer service, all of it overlapping with what these tools can do.

Q: Is everyone in business education taking this as seriously as Wharton?

Everybody is doing things. Maybe two years ago some schools thought they could wait, but right now no school believes this is something they can sit out. Some are making more cosmetic moves, you start a center, you hold a one-day conference, you offer a course in Python. But we’re really rethinking the curriculum. And talk to ten people at the same school and you’ll hear ten different things, which is good. It’s really a huge laboratory now. Everybody’s trying things out, and best practices are starting to emerge.

Q: Wharton launched an AI major last September. Does it last, or does it go the way of the “international business” majors of the 90s?

The biggest policy discussion in the school is whether there should be an AI major with dedicated courses, or whether AI should just be a topic covered in every course. My answer is that it’s not one or the other, it’s both. You need to update almost every course in a business school with some AI content, because the world is changing. At the same time, it makes sense to have some courses fully dedicated to AI. Maybe in the future we won’t need those. In the 90s we had international business majors, in the 2000s it was internet business. Every business became global, every business was touched by the web, and the majors faded. The same will probably happen with AI.

---

Michael Olenick, JD, is a former INSEAD research fellow and founder of VSTRAT.ai. He has spent four decades building AI and expert systems; his research has been used at INSEAD and, before then, cited by Congress and media on how to resolve the 2008 era financial crisis. Michael currently lives with his partner, Anastassia, in Austin, Texas.

© 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.