At Harvard Business School, ‘We Forge The Best Tool There Is, The One On Top Of Your Neck’

Latest in a series featuring short interviews with faculty at various B-schools focused on the use of AI in business and business education. See also Rebuilding The Role Play (Kellogg), Learning Requires Friction (Wharton), The Deck Is Dead (Stanford), and We Teach MBAs To Apologize About The Best Tool They Have (INSEAD).

The first time I heard of HBS, the initials, was just after I’d moved to San Francisco during the first dot-com boom. I’d sold a popular early website, and a young executive asking my thoughts on the future of the web dropped those three letters into the conversation. I had to admit I didn’t know what they meant. Nobody has to admit that now. Harvard Business School is old but not the world’s first business school, large but not its largest. Its influence is a different matter. Every MBA student, professor, and exec ed participant on earth knows two three-letter acronyms: the degree, and the school that invented it.”

Harvard Business School can afford anything. Every first-year student, all 930 of them, gets ChatGPT and Claude; even the classes of three years ago got the former, well before most schools had an AI policy. Now there is a token budget most startups would envy. An internal group of AI architects and engineers, DTX, builds tutoring agents, feedback agents that check a student’s cash flow model down to the cell, and career navigation tools. An AI Academy trained 1,500 faculty and staff last year. AI avatars of CEOs, many of them HBS grads, sit in on case discussions. There are simulators where professors rehearse teaching their own cases before facing a room.

Provenance explains the inventory. As the school that invented the MBA then followed on with the case method, HBS has a unique history. But it does not explain the emphasis. When a school with unlimited resources and unlimited demand can buy every tool on the menu, what it chooses to double down on is a revealed preference, and HBS has revealed one: judgment. The tools are the floor. The product is the thing no vendor sells.

That choice runs deep in the institution. The elite business schools famously send their own doctoral students away, asking them to prove themselves elsewhere before any hope of returning. HBS keeps some of its own. It invented the case method a century ago and still embraces it while evolving it. This is a school comfortable trusting its own judgment about judgment.

Mitch Weiss is the Faculty Chair of Technology and Transformation at HBS, a professor of entrepreneurship, creator of the school’s public entrepreneurship course, and one of the faculty behind its required first-year AI course. Talking with him, two phrases kept surfacing that deserve to outlive the interview: Students must still “forge the tool on the top of your neck.” And in an era obsessed with training AI models, “We still train the human model.”

A final phrase worth remembering was Mitch’s summary of AI’s place in business and business education: “We’re all navigating these waters together.” Our conversation below was edited for length and clarity.

P&Q: What first interested you in generative AI?

Mitch Weiss: It came from two directions. My main area of interest is innovation in government and for governments. I created and teach the public entrepreneurship course at HBS, so I was always researching and teaching about the latest innovation, whether it was autonomy, blockchain, AI. One early AI exercise that’s been quite memorable for folks was built around the phenomenon on Storrow Drive here in Boston, the trucks getting stuck underneath the bridges, and using AI to figure out why and how to solve it.

Then in the fall of 2022, like most of us, it was ChatGPT that enlightened us to what this might mean in classrooms. I was faculty chair of the first-year program at the time, so I became very interested in what it would mean for teaching. It was the intersection of those two things: AI in the public sector, because that was my area, and AI in the first-year program and higher education, because I had some responsibility for what was going on at HBS.

Tell me about the first-year AI course. I haven’t heard of another school doing that.

We have a first-year required course on artificial intelligence. Every single one of our students has to take it. It’s a mix of our traditional case method discussions and a lot of hands-on building. This past year the students had several hands-on exercises, including building multi-agent teams and building AI evaluations, then three build studios and final projects where they had to build an AI system, either heavily with AI using the coding agents we provide them, or one that relied a lot on AI.

It was fascinating to see what they built. But it’s still this mix: we’re going to make sure you’re exposed to lots of AI tools and get quite proficient with them, and we’re also going to make sure you still forge the tool on the top of your neck. Thinking with your fingers and thinking with your head. We’re doing both.

Years ago I wrote a case on autonomous vehicles asking BMW what happens when the ultimate driving machine ultimately drives itself, the same as every other car. What happens to differentiation when everyone has the same AI?

It doesn’t have to be that way. There can be a tendency in some of these tools to produce median, modal, common stuff. But you can also use these tools to produce things that are quite unique. You can induce them to think in new and different ways. People say they can’t produce new ideas, but that’s not true.

We are a school that at the core teaches education for judgment. It’s that judgment and decision-making which distinguishes leaders and is called upon in those big leadership moments. If you don’t use these tools with ingenuity or care or discipline, you could produce common stuff. If you use them with ingenuity, creativity, judgment, you can produce differentiated value.

The strategy field has bifurcated. One camp says AI can make your strategy. The other says the frameworks help, but at some point a human has to take accountability and decide what the firm actually wants. What do you think?

I still believe in educating for judgment. Your judgment, your decision-making still matters. You and your AI together, fine, but not just your AI, at least not for now.

And to the larger point about what business schools are for in an era full of AI: there’s still lots of room for higher education. Humans still produce knowledge, and that knowledge trains the models. We still train the human model. When you’re a manager deciding who should take on a task, you’re deciding between a human and artificial intelligences, so we still have to train the human model. When people are in school, you’re helping them build up memory and context, which are now the coin of the realm in AI, and they leverage that for the rest of time with their tools. You’re connecting them to other people. And at the highest level, you’re teaching them what questions to ask, what goals to chase, and in the end, whether what you did was any good or not. There’s still a lot of room for humans in those things.

How are students receiving all this?

One shouldn’t generalize too much. There are students who are extremely enthusiastic and hopeful, and students who are nervous and worried, and sometimes they’re the same people on the same day, at the beginning and end of the same class. We try to accommodate both the sense of opportunity and the sense of risk.

We gave all of our students access to ChatGPT and Claude three years ago, probably the first leading business school to do it, and we give them more tools on top of those, with quite a substantial token budget. But the purpose of those tokens is not to proselytize about AI. It’s to give students the opportunity to engage with it themselves and formulate their own views. So: lots of enthusiasm, lots of building, lots of founding, and alongside it plenty of curiosity, care, and caution. It’s good that we have a mix.

You focus on public entrepreneurship. What new businesses are you seeing?

In my world, what you see is people, including former students of mine, using AI to solve some of the more vexing problems in government. Permitting: there’s a huge housing shortage around the world, so building permitting is a huge area of interest. Procurement: massive amounts of money spent by governments buying things, and plenty of AI companies seeing whether they can help do that more efficiently. Government workforces, public safety, transit and mobility, there are AI-native companies building in each. I wouldn’t say new areas so much as new approaches to old and vexing problems, a thousand times over.

When I’m asked whether this series is positive or negative, I say there are really only two negatives anyone names: students worried faculty are behind on AI, and faculty getting thirty identical AI-written essays. Has that come up at Harvard?

We happen to have this technology that’s more than 100 years old called the case method, which many of us believe is perhaps the best technology for teaching in this era. It was always meant to invite students to think, to engage their critical thinking, to cultivate their judgment. I don’t want to be head in the sand; obviously we need to make sure students are doing that and not shirking with these tools. But sometimes people say, for example, you really should move to oral exams. Well, we have oral exams in the classroom all day, every day. That’s what we do.

So sure, we’re mindful of it. We desperately want students cultivating their own critical thinking, and if we hone and improve our own case-based, Socratic teaching, we can really help students develop this tool on the top of their neck. We’re also experimenting with other ways of engaging students: AI-based simulations and role plays, tools that make students pull out information rather than have it handed to them. We’ve built AI teaching simulators so teachers can practice teaching their cases, and tools to help teachers evaluate their cases and improve their teaching plans. The case method can be really robust in this moment, but it’s incumbent on us to become better and better teachers.

The case method was born at HBS and it seems to be evolving fast. Where is it going?

We have the privilege, and honor, and duty to help teachers everywhere use a method that may be very robust in this moment. People worry: I can’t just ask students to regurgitate something they read, I can’t just assign an essay, I can’t just lecture. Well, we have this method, and we’ve had it a hundred years, and it has evolved with every other technology. It wasn’t always an eight-page PDF. At one point cases were bound books. When film came out, they experimented with film, then TV and multimedia. We’ve constantly been evolving the artifact that is the case while doubling and tripling down on teaching critical thinking and judgment through the case method.

I’m involved right now, and I don’t want to say too much about it, in an experiment around essentially cases as markdown files. What if you made the case native, so the tools could ingest it, token-efficiently, and engage with the student immediately in ways the case writer had instructed? It could invite you into a role play immediately, invite you to analyze data, invite you to read the raw transcript of an interview and decide what you make of it.

We’ve had faculty give students the chance to interview the CIO or CMO or CFO of a case. We’ve had faculty bring AI avatars of a CEO into class to sit there during the discussion, and when the students aren’t contributing enough, you can turn to the avatar and ask what it thinks. So it’s both. Our dean is a big proponent of “and” thinking. The written case still matters. We don’t want to abandon having people read, or frankly, write. It’s and: on top of that, the role plays, the avatars, the raw interviews.

My most popular case is the Marvel turnaround, and the executives who lived it told me the whole thing came down to about five people. Is the AI transformation at HBS a five-person story?

I thought that’s where you were going, and at this moment I’d say it’s a lot of people. A colleague of ours, Tsedal Neeley, founded what we call the HBS AI Academy, and we trained 1,500 people at HBS last year in AI. It’s a whole-of-organization approach: everybody learning through cases themselves, learning through hands-on experience, much of the same curriculum our students had. The staff went through it, and many of the faculty.

What else should readers know?

The biggest opportunities we’re working on are about using AI to extend the experience of the MBA. Before you get here, while you’re here, and after. Converting what you learned into context and memory you take with you, elongating the learning experience with these tools. And what we call cascading: you come to HBS as an MBA or an exec, how do you go back into your organization and your community and help them learn this? Lifelong learning, and cascading learning into the organization.

And we’re fortunate to have an ecosystem behind it. The HBS AI Institute leads on AI research, teaching, and impact inside companies. DTX, our digital transformation group, is AI architects, software engineers, and product managers who built our early tutoring agents and our feedback agents, which will give students feedback on the models they built, down to the cell, and agents helping students navigate career searches. And a really strong IT group that deploys these tools across 930 students in the first year, another 930 in the second, and all of our execs. Besides the students and the faculty, we’re blessed with an amazing staff team.

---

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.