This series is on AI's disruption of entry-level work and what universities should do about it. Give us a quick picture of what you have going on right now that ties to that question.

So I'm doing a lot of research right now for a new book, which is looking at what strategy really means in the future. And increasingly, the theme is that more and more of our world is becoming dematerialized.

And what that means is a couple of big implications for young people. Firstly, that it has never been cheaper or easier to start something of your own. So what I'm seeing, these new tools are able to give young people, is you can actually create something of value without it having to be the product of a large corporation.

If you think about it, why did we need big companies? We needed big companies because they had assets and resources that individual people just couldn't afford or couldn't assemble. It was too difficult. With today's tools, you can have individuals accomplish astonishing things.

So there was recently a young man in Israel who spent six months by himself coding an application that proved to be so desirable that Wix bought it for $80 million. Six months by himself.

So there's a lot of tearing of hair and rending of garments about the lack of employment, traditional employment at the lower end of a large corporate hierarchy. But if you think about it, I know young people today that are doing their own thing. Almost all of them have side hustles of some kind, maybe not even a hustle. Things maybe we used to call hobbies can now be turned into actual economic engines. We've got people doing influencing. We've got people creating small apps. We've got people coding things that never used to be possible to code before.

So I think there's one bucket of work which really looks at, how is the changing nature of economics changing what's possible up and down the career ladder? And if you don't need a big company, if you and 10 of your friends can do what you guys are doing, create your own company, think about scaling it without needing all the apparatus of a large corporate bureaucracy, why on earth would you want to have one?

So I think there's one piece of research really looking at this world where the value creation is becoming much more distributed than it ever was before. Now, one of the knockout implications for that is that our classic ways of financing startups may no longer be relevant. And I've written about the fact that I think venture capital in its traditional form, and I would date that from the post-Netscape IPO to right around today, may not be necessary anymore. If you're a startup person and you can create a viable business, and maybe it's not going to be the size of Amazon, but you can create something that's viable, and you don't need VCs sitting on your cap table and telling you what to do, why on earth would you want them?

So I think there's a whole lot of second-order effects that start to happen when you make something that was challenging like that very easy.

So I think it's useful, perhaps, to reflect on the nature of disruption. And you'll be familiar with my work with Clayton Christensen, somebody I worked with a long time, and he used to joke that while he was very successful in getting the idea of disruption out there, one of the problems is, it's come to mean kind of anything that's a big change. And I think it's more useful to look at two very specific qualities of what makes something disruptive.

So the first quality is, it makes something that was once really hard, easy, so that many more people can do it. And the second quality is, it makes something that was really inaccessible or expensive, widely available. And that has two effects.

It destabilizes the older order, because what you might have depended on as a competitive moat, for example, you no longer can. And what you might have once just assumed would protect you: if I'm making steel plants, it requires millions of dollars of investment, so it's not as though any Tom, Dick and Harry can just get into the business. And once you have this shift towards a dematerialized economy, where more and more value is being created out of software, IP, intangible assets, a lot of those moats disappear.

Now, for an incumbent, that's pretty terrifying, but for a startup, that's liberating. What do we know? We know it creates huge drivers of demand, because now something that people were basically locked out of, they can now be in the market for.

On the HBR Strategy Summit podcast, you said the right move on AI is not to write an AI strategy, nobody had an electricity strategy, but to give people the tools to experiment. Please expand on that.

That's a great question. So if you go back to the earliest days of the introduction of electricity, it has a lot of parallels to what we're thinking about with AI today, because there was a utilization component, there's a whole infrastructure component, there's a standardization component, there's a safety component, and we have still not seen a lot of that necessary ecosystem created for AI products. So let's just put that as a preamble.

But before there was reliable electricity that you could count on, all factories were designed in a linear logic. And the reason for that was, they were all dependent on a single power source. So a river, a steam engine, some kind of single source of power. And the skill and the capability of building the factory was all around, how do you maximize the output of that factory, given this rate-limiting step, which is single source of power.

Once you introduce electricity, now all the machines can operate independently. And so it just completely relaxes that old constraint of how a factory needed to be designed and opens up just vast new opportunities for greater productivity. You could separate out different kinds of work, you could turn the machines on and off, you could have them interact with each other in radically different ways.

And yet, the old regime took a really long time to give up, and there are lots of reasons for that. So one is that you had a lot of capital invested in the old regime. So to get rid of it, throw it away, you have to write all that down. So depreciation is real. Secondly, there's a skills component, similar to what we're seeing with all the dread and fear around AI today, which is, hey, I don't know this thing, I'm all about linear factory, I'm not about independent machines. And so there's a human resistance. And the third thing is, there was a lot of learning that had to take place before people really learned what electricity could do.

And I think the parallels to AI today are, yes, AI is going to make a whole bunch of stuff we used to invest in no longer particularly valuable. Second, a lot of the people that are in power don't have comfort that they know how to operate with this new technology. Third, we're just at the very early stages of learning what AI can actually do, what it can't do, where to use it, where not.

And so if I come back to your AI strategy, that's as silly as saying, what's your electricity strategy? Because your electricity strategy depends on your business strategy. It depends on, what am I trying to make for whom, what's going to be my competitive moat, what will protect me? And so my argument is, figure out your strategy first. So what are you centering your company on? Then work backward into where could AI be helpful. And then back into your conversation about skills. Okay, then what kind of skills and what kind of people do I need to be supporting those efforts with AI?

And I think the companies that are getting this right are doing that. So a great example of this would be Shopify. Shopify makes websites and supports smaller businesses, typically makes it easy for small businesses to do e-commerce. And the Shopify CEO has basically said, look, before you ask for headcount, or before you ask for budget, give me a demonstration of why you can't use AI in some way to do this thing you want to do.

So I think what's good about that perspective is, he's not saying no. What he's saying is, learn enough about the AI to prove to me that it can't do what it is you're needing. So he's kind of creating a glide path for people to learn about it.

Then once they learn about it, what he then looks at is, he says, well, okay, how can the AI now be used in service of our customers? And so as an example, they have a product where, if you're a Shopify customer, and you have a small business website, and let's say you make chocolate or something, and Valentine's Day is coming, you can actually speak to the AI and say, oh, I want to change my website so that it has cupids and hearts and music that's appropriate for Valentine's Day. And the AI will go off, within five minutes it says, okay, Valentine's Day website, here's nine or eight different templates you could choose from.

Pick the one you like the best. You pick the one, and it goes and reformats your website, and you don't have to do anything other than talk to it.

Now, why I think this is such an interesting example is, previously, if you're a small business person, and you wanted to make a change like that to your website, it was an ordeal. You had to find a designer, figure out what the design should be, get someone to program the mechanics, usually it'd be wrong, there'd be bugs in it, or something wouldn't be right. And it would take days and it would be expensive. And so, again, something that was hard to do now becomes easy and something that was really expensive now becomes affordable. And so Valentine's Day comes and goes and now you want St. Patrick's Day, you tell the AI the same thing.

And so that to me is a really nice example of how you're taking the friction out of a system that used to require lots of different interactions. So if your strategy as Shopify is to make your life as simple as possible, to make e-commerce as easy as possible for your customers, then that use of AI makes perfect sense. Now, if your strategy was something else, having a website you could talk to and it would reconfigure itself may not even be relevant.

Novartis reorganized itself around curious, inspired, unbossed, and Vas Narasimhan treats senior versus junior as an irrelevant concept. What does unbossed actually mean at the entry level? What is a graduate from the class of 2027 doing in their first 90 days in that kind of organization?

One of the things Novartis loves most about unbossed is, nobody knows what it means. And so it gets a conversation going, which is what they want. They want people talking about it.

What unbossed is all about is being able to take initiative, which is aligned with the company's purpose, without having to ask permission. So it's about using agency. And I know they've drawn very heavily on Dan Pink's theory of human motivation, which basically says what really drives people is agency, so a sense of control over what you're doing, mastery, which is learning, and that's where the curious part comes from. And then a sense of connection to a greater purpose, which is where the inspired comes from in the Novartis language.

So when they say unbossed, what they mean is, let's create the conditions in which we, in connection with our colleagues, can make progress on important goals without having to ask permission, cover our tails, make up stories. If we make a mistake, let's be open about it, let's learn from it, let's be supportive and psychologically safe in our conversations with one another.

But if the strategy of the company, which is about reimagining medicine, innovative medicines that have never been seen before, if that's your strategy, you need a ton of human creativity.

And that's where Vas Narasimhan, who's their CEO, struck on the concept of unbossed. He said, if I want people to be creative, I can't have them behaving like little soldiers all in a line. I need them thinking broadly and thinking more creatively.

So for the new graduate, what that means is, get to know your people, get to know your network, figure out where you can add value and go ahead and do that without having to ask a lot of permission from people.

I want to push back a little bit on that. My main profession is sales, and my first sales job was more of a military boot camp than it was a job. We were told exactly what to do. We worked 13, 14 hours every single day, and we were doing the most boring grunt work you can imagine. Over and over and over again until it was in our bone marrow. I appreciated it, but the work itself was extremely non-charismatic. What's going to happen to that now? Because most of the work I was doing is gone now. My main job is getting eaten up, which is both a blessing and a curse. So what does unbossed mean for a student that wants to graduate and learn something, have an apprenticeship?

Well, I don't think the apprenticeship model goes away. I mean, if you think about all of human history, and let's go way back, before we had the Industrial Revolution, which by the way in historical terms is relatively recent, how did human beings learn anything? Well, the older and more experienced members of your tribe would show you how to do stuff because the survival of the tribe depended on eventually you learning how to catch a fish, or score a pathway, or escape from an animal. And so it was always about senior people passing on to junior people key skills.

Now what I do think changes is this sort of crucible that we typically have put young people through in professional jobs. And there's actually a bunch of research on this which shows that, for example, in investment bankers, consultants, financial advisors, that there's this sort of hazing that they go through, where it's the intense hours, it's the non-stop stuff, and a lot of the work itself is not particularly value-added.

I would suspect that kind of crucible, and I do think it's essential, by the way. I do think there is a time in your life when you just absolutely need to be pushed to develop to your fullest potential. And I do think of it like a crucible. It's that moment of pressure that creates the polished diamond. But I think the nature of that work is likely to change.

I think you may well see much more intense partnerships between more knowledgeable senior people and less knowledgeable junior people. There's a guy called Matt Beane who's written a great book on this, called The Skill Code, which I would recommend you look at and talk to him.

But one of the things he talks about is, a lot of what we have sort of taken for granted and unconsciously accepted is that, how do junior people get developed? It's through these kinds of jobs, where you get stuck in a situation, you're told what to do, you fill out the forms. It's boring as hell, but it helps you learn as a group what you're supposed to do, and that repetition develops that muscle and that thing. Well, we do that without really planning it.

And so, what if we did that task, that job, of developing skill in young people but much more deliberately? And so what I'm seeing firms doing now is they're gamifying their workplaces. So there's a manufacturing plant, I can't remember where it is, but they've designed a simulated plant of their own, because it's all run by machines and robots. It's very high-tech.

And what they set their young people to do is working in the simulated space. They had them play games with the plant. So you arrive, and your training is, you get a task or an order or a customer requisition or whatever, and your job is to respond to that in the simulated plant. So you're not working with the real factory, but in the game. It's like a flight simulator for pilots.

So you're practicing on the plant. And as you practice, as you spend your time getting the hours in the simulator, you start to realize what the interconnections are, how a change in one system affects something over there. And then eventually you graduate to being able to work at a more junior job in the main plant. And then eventually, as your skill develops, you get better and better.

So I don't see why that wouldn't be a model that firms use. And it's always frankly been strange to me that when it comes to athletes, when it comes to pilots, when it comes to Navy captains, when it comes to fighter pilots, we have all these mechanisms to train them, because we understand making mistakes in real life is really problematic. So athletes get filmed and then they go over the films and then their coaches give them guidance on what they did and how they could do better next time. And next time they do better. And pilots have to spend time in flight simulators, hundreds of hours in flight simulators, to make sure that they've got the skills before they do the real thing.

And yet when it comes to some of the most important jobs in the world, management jobs, where your decision could affect hundreds of thousands of people if you're senior enough, none of that infrastructure. We don't let you simulate it. We don't film you, we don't give you a coach afterwards. I mean, maybe some companies do, but I would say they're probably rare. Most of the time what happens is, we take people and sort of randomly put them into these roles and hope that they can succeed. And maybe there's a little bit of scaffolding about coaching or somebody who's helping you, whatever, but I think we could be a lot more deliberate about it.

There's another great book you should look at. It's by Tomas Chamorro-Premuzic, and the book is called Why Do So Many Incompetent Men Become Leaders? And his central thesis in the book, not to give it all away, but his central thesis is, as humans, we tend to confuse competence with confidence. And so people who are very confident, even if they're no good at anything, tend to be seen as leaders, rather than people who just kind of get on with their lives.

From where you sit, what are universities too slow to change right now? Is it possible for you to dissect and name the slowest part?

Well, so first of all, universities are big places. So I think you need to think about which activity of the university. Are you thinking of undergraduate education? Are you thinking of graduate education? Are you thinking of business schools? So I think it's a little bit different for each.

For undergraduate education, I think one of the big challenges is that for the student, the sort of key job to be done, and Clay Christensen would have used this phrase, that's the coming-of-age job. We're 18 to 22, we want to learn who we are in the world. We want to have our thinking challenged. We want to be taught to think critically. We need to be able to have those critical feedback conversations.

And depending on what kind of school you are at, that may or may not be the experience you're being given, in that sort of ability to really think critically and ask good questions and probe for errors and consistency, that kind of thing.

So I think one of the things universities have not sort of aligned on is, what do students need to be able to have an impact on? That doesn't actually mean just spitting back whatever they read in a textbook verbatim, because that's not what learning's about. We inherited that idea from the industrial age, where basically what you didn't want was very educated workers. Let's be clear on this. What you wanted was people who took a pencil and moved it from there to there to there to there all day long and didn't have the imagination or the brains to be bored.

So a lot of our education system in the West is not actually designed to equip people to be disruptive critical thinkers. It's designed to create robots in factories. So I think the first thing that we really need to grapple with is, what are designs that would promote impact generation rather than rote learning?

And an example of a school that's doing this really well is the Minerva Project, which is out in San Francisco, run by my friend Ben Nelson. He was the founder of a university. It's called the Minerva Project. Minerva University is part of it, but there's a lot to it.

But one of the things Ben comes in with is, he says, look, the design of the university that we have, certainly for undergraduates, really comes from the age of monks in monasteries, and very few people could read. And so, if you were the person who had all the learning, your learners were sort of gathered around you to absorb all this tacit knowledge, and then they too could go spread it on. And we've sort of inherited this model that hasn't really changed since the day of Socrates.

And Ben's idea is that the students should be able to take their learning, and the learning itself should be fairly standardized. So this idea of every professor inventing their own curriculum, I mean, why? If you know there's basic things you want to learn, then why not make that standardized so that everybody has scalable access to that same set of ideas. And then what the professors in Minerva do is they help the students apply that learning to a meaningful set of outside activities. So that could be a project. It could be some intervention they want to make in a community. But they're doing something with the learning rather than just memorizing.

On that podcast you also used the loading dock test: ask leaders what would happen if somebody on a loading dock had an idea, and you get blank stares. Adobe has Kickbox. What is the equivalent of Kickbox inside of a university? What mechanism should a 21-year-old senior with an AI idea have available that almost none of them do?

Well, I don't actually think that's true. Universities, at least the big ones, the big research universities, they're, on the one hand, incredibly stodgy bureaucracies, and my place is more than 250 years old. We're older than the United States of America at Columbia. So on the sort of big structure, it's this incredible sprawling bureaucracy, but within it, it has all these pockets of innovation. So we have centers, we have study groups, offices, we have people that get grants, we have people that have their own little research units.

So I would say universities are better than corporations when it comes to somebody with a bright idea being able to find like-minded souls. And as an example right now, I'm starting a series of conversations with people about how do we think about what society we want when this AI bubble that we're in right now crashes? Like, how do we begin to think about the pillars? And I'm pulling together people from behavioral science, from public policy, from economics, from whatever, and they're easy to find, they're easy to respond to.

So I think for a young person, one of the wonders of a university is that there's all those people doing all this fascinating stuff all around you. And since they are dedicated to learning and discovery, most of the time you'll be able to find somebody to talk to and work with. Now maybe not the Nobel Prize laureate, that is very, very busy, but perhaps his assistant or her assistant or somebody in their research group to talk to. So I think it's actually easier in a university than in a corporation typically.

Last summer, writing on the Gen Z crisis, you said that writing code can be delegated to machines, but soft skills cannot. Can you be specific about those soft skills? Which of those should the class of 2027 walk out with that they currently do not? And what is the school actually responsible for teaching them?

Where do I even start? So I think soft skills, and again, this is where universities perhaps have an opportunity they haven't leveraged.

If I go all the way back to my high school days, we had two kinds of classes that young people took. And this is before they even get to college. We called it Home Economics and Shop. And in those classes, what you learned was life skills. You learned how to sew something. Now it doesn't mean you were going to become a seamstress, but at least you had enough exposure to know what it looked like. You learned how to cook stuff, you learned how to balance a household budget. And in doing all that you were working with your peers, so you learned how to bond with other people, you learned how to cooperate, you learned how to delegate tasks, you learned how to keep each other accountable, and on and on.

And so I think in universities, we don't have very many courses, I would say, that are the equivalent of that. But I think the soft skills that we're looking at are things like, how do you make an empathic connection with someone? How do you create a bond with someone? How do you ask for a polite favor if that's necessary? How do you read the room? How do you see who's uncomfortable, who's not on board, who's possibly scared, what people's emotional states are?

So there's this whole kind of surround of human connection and the ability to navigate that, which I unfortunately see a lot of young people either short-changing or really not giving credence to. I mean, just go into a restaurant and look at Gen Z people together, and what's there? It's bloody phones everywhere.

So you're not really present, and I'm not trying to be like a get off my lawn, kids kind of thing. But I think these things have become so ubiquitous that we've forgotten that you can have a much more authentic present conversation if you put the thing away and still don't glance at it every 15 seconds.

I mean, I think what we've forgotten is, these devices are designed to be addictive. And the thing about learning human skills is, humans are messy. They're often boring. They can be really unpleasant, they can be irritating, they can get on your nerves. The phone's much more predictable and it does exactly what you tell it to and you know what you're going to get. But in just pursuing that, you really miss that ability to create a human connection, which is irreplaceable.

There's a question that I've always thought about because I keep hearing, you need to develop your social skills and empathic skills. But then I've also thought about, what about neurodivergent people who might be wonderful, incredible, intelligent human beings, but in the room setting, in a group setting, might not be able to show those things? I've always been afraid that there might be some sort of elitism created where if you don't fit in, you don't get to partake. Do you feel like that is a worry?

Oh, it's a huge issue. And it's something that Seth and I are actually actively working on. We're working with a company called Ultranauts. They were founded by an MIT grad who realized that we have all these people on the autism scale, neurodivergent, very capable, very bright, but they were systemically being shut out of many opportunities just because, as you said, they didn't have social skills. They didn't have the cues to let people know how capable they really were.

And so this guy decided to found a company to take advantage of just exactly those skills. Their first line of business was doing quality assurance testing for computer code. Now what's super interesting to me is that non-neurodivergent people actually can't do that very well. So what happens, if your brain is wired the way most are, is, it will fill in the blanks, without the correction being in reality, it will correct the errors and so forth.

But somebody who's neurodivergent often is able to break that. They look at it and they go, no, no, no, it's C, and then there's missing something, and then there's a T. There should be an A there, and it doesn't fill in the blanks. And so what they realized was, there's this whole need for people who could really do this exceptionally well that wasn't being met by the workforce as it stands.

And so what they started to do, and it's marvelous, is they've created this whole set of activities which bring people who are neurodivergent into good jobs, where they can build a life, where they can be able to function and have their best flourishing in environments that are suitable for them. So we're seeing a lot more action around that, more activity around that.

And, to be honest, a lot of the big tech companies, a lot of their most talented programmers are a little kind of on the spectrum. So there's a lot of value that's wasted if we don't really take that talent into account.

You and Ian MacMillan published Discovery-Driven Planning in HBR in 1995. Steve Blank credits it as the source of Lean Startup. Your method became the operating system of how a junior person learns under uncertainty. AI now does that learning in seconds. If you had to bet on what the operating system of the 2035 business graduate looks like, what is your bet?

Well, I think the centrality of learning is not going to change. And if you learn quickly, I think the principles behind Discovery-Driven Planning are not going to change. The basic idea is that when you're facing high levels of uncertainty, and we're facing high levels of uncertainty, you can't really plan as though you had a platform of deep knowledge to go forward on. You have to break your learning down into chunks, which I call checkpoints, and then test your assumptions at each checkpoint.

Now, where I think AI is going to be incredibly powerful, and I'm actually working on software to make this easier and easier and to make our learning faster, is I think you can iterate much more quickly using AI than we were when we had to do spreadsheets and pen and pencil kinds of things.

I think you can also learn from past experience much more easily. So I always encourage companies to do, you know, there's a project that got shut down, or even if there's something that went well, do a post-mortem, write it up so that you can share the learning. Half the time, more than half the time, they don't. Why? Everybody's busy, everybody says, I don't have time. But if you had the record captured in an AI and you went to your AI system and you said, I'll just make this up, I want to do a robotic pallet sorting machine, and then the AI could go back in its memory and say, ah, you know, we had a robotic pallet sorting machine idea a few years ago, maybe there's something there for you to learn from.

So I think from a learning perspective, it could be much richer and much more complete than the way that we've been doing it so far.