Preparing for an AI-Driven Future in Business – Wharton in Focus: New York

So I'd add a third element, which is R&D. I think organizations have to invest. If there's ever a time to spend money on R&D, it's now. And organizational R&D, stuff we don't typically do. You're going to have to have teams trying new techniques for using stuff that may not work. And you have to be running experiments that might fail. That is a cost, right? You're going to pull your best people off the line to have them do new things. And, you know, I've been hearing great Wharton alumni talk about some of these experiments, right. And so that is part of what you need to be able to do. So in terms of allocation, just adopting without changing anything isn't going to get you very far. Right. So I think that you need all those elements together. We don't have a clear model for what adoption looks like. And if it's just adoption at the individual level, you'll end up having bottlenecks inside your company. You don't want a thousand PowerPoints. So I think you need to, like, it's management attention as well.

I wish I had all the answers. The four things I tend to focus on are deep knowledge, wide knowledge, taste, and agency. Right. So deep knowledge is you should know a field deeply enough that you can have judgment calls about this. Like it's great, like, AI agents are incredible if you have enough of a sense of what you want them to do that you can assign them tasks and when they come back you're like, that's not quite right. I know how to fix it. Like, if you're in that shape, you will be the most productive you've ever been in your life, because your deep knowledge lets you make quick judgment calls rather than spending an hour reviewing every AI answer. Wide knowledge because now you can do things you couldn't do before. Boundaries have changed within firms. Like we, I don't want you just now accounting. I want you to know about accounting and management and psychology. And the more you know, the better off you are. Right? So those are two things we do at Wharton, deep and wide. I think the other piece is agency. Right. Which is what we try to teach in entrepreneurship. Like you can do more things now than you could before. So what do you want to do? If you're just waiting for instructions, which is something I don't think we typically associate with Wharton, but a lot of people just wait for instructions. I think you're in trouble. And then taste, like, you're gonna have to make some judgment calls about what works and what doesn't. I asked you to pick which one of those ten ideas you liked best. Those become consequential choices when you can just turn those ideas into businesses. Having a good sense of taste or your own developed identity can really help.

So the easy stuff is talking about how you should, your kids you use AI today, the hard stuff is what do we do next? And I wish I had like an easy answer where I'm like, oh yeah, I know. Nobody like, every bet you make that's over three years from now, or even the next few years, is about how good you think AI is going to get, how fast. Like that's what that bet comes down to. And nobody has any answers for you, and nobody even knows what it means for the job market for this to happen. So I tend to tell people, you know, this seems like, you know, I did a lot of work with Matthew Bidwell, who some of you might know is a career researcher, and a lot of the work that comes out of the career work, and Peter Capelli and people like that, a lot of that work is about how careers are long and change a lot. So I think resilience is going to be key. The job you're going into was never going to be the job you're going to end up in. That's even going to be more true in the future. So I have a kid who wants to be a doctor and a kid who wants to be a lawyer. Those are, in one way, jobs that are really targeted, but they're also jobs that are very diverse with many different kinds of things you do. Of course, the other option I just tell people is just go into a regulated industry. That's the other choice.

Okay, so the craziest thing is, just to show you how far reinvention can go, there is a company called Strong DM that pioneered something they called the software dark factory. And so if you know about a dark factory, the whole idea is like, if I have a wholly robotic factory, I don't need lights because there'll be no human in the factory. There's this software dark factory using AI agents. They had two rules. No human can ever look at the code, and no human could ever touch the code. So the goal is that you have to spend $1,000 on tokens and AI cost per engineer per day, and the system would just churn out, not only churn out the software, but then it built its own fake Slack and Salesforce and everything else to test the software, and it would have put personas in and run through it, and then it would churn out software package after software package that was ready to go. Right. So the most radical version is what do you do, you just direct a thing to go, you turn the machine on and it makes things happen.

The, the, I mean, they were getting like, they invented this and within a month it was up and running.

Yeah. So all of you, if you're in the right place, you're like me and you have imposter syndrome. You should. Like everyone is making this up right now. We have no playbook to run. There's no consulting company you could hire who knows how to tell you how to do AI implementation. There's no one example. It's an experimentation process. I find that the most important capability is actually leaders who are willing to take a risk. So taking one of our own great alumni, Nikolai Tangen, who I've talked to about this before, he, you know, he runs the Norwegian sovereign wealth fund, the biggest pool of money on the planet. He personally overcame all of his risk managers to say, we need to get AI in use. Whenever you go to a meeting with him, he asks you how you use AI to do something. He put AI training in not because he thought the training was that useful, but because it made people put hands on keyboards. When I last talked to him a few months ago about this, he said 50% of his people are writing code now, right? Only 20% of his office is coders. Like, there is a transformation. Leadership is the key to making this happen. Leadership loud and proud. You have to set up rules for making this happen. It can't just be AI gets implemented and something occurs or, and if you just say increase productivity, congratulations, you're going to get a ton of work slop. Like everyone is just going to produce ten times more generic memos. That's not how to make things happen. So this is more than ever the time for MBA leadership. What processes are you actually trying to do? How are you trying to accomplish it? How do you rebuild your organization? Those are the, those are the defining questions.

Okay. So I would give you the same advice I give CEOs about this stuff, which is, you know, first of all, you actually do need an R&D center. There needs to be a set of people who are kind of doing work and presenting you options. So is that, it doesn't have to fall on your hands. And you know who those people are. They're annoying people like me who have already been coming to you, and there's tons of people in your organization who I talk to all the time who are like, "I'm using AI to do this stuff." Get them in a lab and get them doing work for you is one piece. You need that advisory role. You also should be doing something that is impossible or enters a new market. Try something completely new. We've talked before. I think that there's going to be a huge crisis. We haven't even talked about this. Inside organizations, right? Because I teach people to be generalists at Wharton, especially for undergrads, right? You become a specialist by working for you guys. They become your, your, you know, your interns. And if you've probably noticed, right, we teach you -- So how do they become experts? The same way we've taught experts for 5000 years, which is apprenticeship. And that all broke the summer because every intern is using AI to give you results. The most depressing thing I heard from a CHRO actually was them, they said, "Oh, these AI native kids are so good." I'm like, you're not talking to an AI native kid. You're talking to Claude. They're just pasting every question into Claude and handing in the answer. That's not an AI native. That is just Claude. And so like, that isn't right. And like, and all your middle managers are giving tasks to the, to interns who sometimes cry when they could just give it to the AI. Right. And so nobody's learning anything. So I think one advantage, one thing we should be thinking about is, how do we extend what we do? So I'd be doing something expansionary. And then I'd be also thinking about, how do we get a few success cases inside the organization? I've seen some very large companies, some of the Fortune 10, experiment with pulling their best engineers off the line, putting them with their best subject matter expert, and saying build something impossible in two weeks, right? And they replace an 80-person team doing the work. So I'd be trying to do both that sort of gradual change management that we always teach, expanding in some radical way, and also trying something really experimental inside.