ALISON BEARD: Welcome to the HBR IdeaCast from Harvard Business Review. I’m Alison Beard.
Starting today, we’re bringing you a special series of interviews with some of the world’s leading tech CEOs and founders. We’ll hear their perspectives on artificial intelligence and other critical issues from sustainability to change management to talent. Look for a new special episode of this Future of Business series dropping in your Feed for the next three Thursdays through November 21st. To start, we’ll hear from Reid Hoffman, one of the most prominent and recognizable voices in Silicon Valley. He created or co-created several successful businesses, including PayPal, LinkedIn, where he served as CEO and is now executive chairman, and Inflection AI. He’s also an active early-stage investor at Greylock Partners, host of the Masters of Scale podcast, author of the forthcoming book Superagency, and a sought-after expert on everything from AI to the American political scene. We invited Reid to our recent virtual future of business conference to talk about why he advocates for certain issues and candidates, how he manages technological change and innovation, and what a world transformed by generative AI will look like. He was interviewed by HBR’s editor-in-chief Adi Ignatius, who also shared a few questions from the audience. We hope you enjoy their conversation.
ADI IGNATIUS: I want to just start with the situation we’re in. I think most of us feel this is a tumultuous time in terms of technological change, politics, civility, you name it. I’d love your sense on how business leaders can best respond to this moment of uncertainty.
REID HOFFMAN: So, obviously, it’s complicated in its different circumstances, but I do think that one of the things that’s really important for business leaders to do is to think broadly about what their responsibilities are as leaders of this business because businesses are not just these little abstract profit generation machines. There are things that live within networks. They live within networks of customers, networks of employees, networks of investors, society, etc. So, when you’re a leader of the business, you’re responsive to all these networks. And that’s, of course, part of the reason why the classic dictum in business is to be nonpartisan, to be broadly inclusive in lots of different human vectors because you want to be inclusive across all these networks. Now, that being said, I think there’s a number of things that are good business articulations. When you’re a leader, you say, “Look, this is the kind of society that’s better for my operating my business, the health of my employees, etc.” So, we should have stability. We should have unity. We should be adherent to genuine truth, not the claims of truth, but actual real truth. That kind of thing is really important to state to those virtues where you’re saying, “I’m not trying to be partisan in this.” Now, there are some areas where people will claim that you’re being partisan when what you’re being is rule of law, stability in society. And for example, someone could claim you’re being partisan by being anti-eugenics or being partisan by being anti-Nazi. You’re like, “Well, that’s not partisan. That’s just truth and the right thing for society.” And I think that’s the kind of thing to do as a leader, and it’s what I’ve been trying to do. So, for example, it’s interesting on LinkedIn. I’m very careful to post only the things that I think are the business perspective of what it is. A bunch of other things about my political views I post in other places, and that’s how I sort even my own identity between these.
ADI IGNATIUS: Yeah, so it’s interesting. So, you’re advocating, in a sense, I think, that there are issues, fundamental issues, that every business has, and they won’t be the same from business to business. And on those issues, maybe you shouldn’t be silent, that you need to speak up. And I asked it just because I think there’s a fatigue, I guess, about … and people have gotten their fingers burned trying to wade in with what they feel is a fundamental value, and then it just blew up in their faces. So, it’s a minefield, and it’s interesting that there are some people like you who dive into it, and there’s some people who really feel the safest place is to have their head in the sand. And I suppose you’re talking about a middle road, which is you need to be engaged when it’s really fundamental to your business, to your institution’s values.
REID HOFFMAN: I would change the metaphor slightly. I mean, I always … like driving down the middle of the road, and you get hit by cars going both directions, although it might be actually an appropriate metaphor in this circumstance, but it’s more foundational. It’s the question of, look, if you are being a leader, you are being … and this is one of the things I think is really healthy and great about American society, is that business leaders are leaders. Our leaders are not just political leaders. Our leaders are not just military leaders. Our leaders are not just university leaders, but they’re also business leaders. And actually, I think part of the health of society is fundamentally in the health of business. So, articulating what kinds of things are good for the health of business, and I tend to find business leaders who say that the only really relevant variable is corporate tax level or tax level are clearly very narrow-minded in what they’re … the only variable. It is a variable, but it’s not the only variable in the health of society. And so it’s important to have the discussion but to have the discussion with thought and civility. I’m making the argument for what is a better case for business. You can make an argument back, but it’s a civil discussion.
ADI IGNATIUS: So, let me switch gears a little bit. You have had an incredible career in Silicon Valley and driven, in part, by being really good at discerning weak signals before they become obvious to the rest of us. Is there a method to this? Is there anything you can impart to this audience on how to do that better?
REID HOFFMAN: There absolutely is. And what you do is you think about … I mean, you start with the Wayne Gretzky line, which is things are changing and skate to where the puck is going, not to where the puck is right now. And so you should start with asking yourself that question, which is, okay, so what are the ways in which things could change or seem to be changing? And then, what are the consequences of that? And if you say yourself, you may generate some of the ideas, but then go ask people in your network. So, it’s a classic thing that I ask when I’m talking to technologists is what do you see is going to happen in the next one to five years in this technology? What do you think might happen with it? What do you think people think might happen that’s actually not going to happen, it’s going to happen differently? And then you begin to think about, okay, which of those things track to what actions you should take differently? And part of it is it’s always probabilistic. Being overly certain is wrong, whereas thinking, okay, there’s a good chance of this. So, it’s a good chance. What does that mean I should do now? Because very rarely it’s like I should presume 100 percent unless I can just correct from it, but it’s like, oh, I should do X. Now, some of that’s investing. Some of that’s deploying technologies. Some of that’s guiding an organization to what you’re doing. And so, for example, it didn’t surprise me that the general press discourse around AI was, “Oh my god. It’s going to change everything. There’s going to be a total shakeup.” And people are generally more pessimistic initially in a disaster. And it’s like, “Oh. Is all this AI thing is doing is how you help you cheat at homework?” And it’s like, “No. Today, with these AI things, literally anybody who is a white-collar professional or is doing anything with language, it is a tool that is useful today. And if you haven’t discovered how it’s useful to you, that’s because you haven’t tried enough.” Right? It isn’t because I went, and I said, “Write me an investment memo. Oh, look. It wasn’t very good.” Okay, try other ways. What should be the due diligence? When I upload my investment memo, and I say, “What should be the due diligence questions that I’m asking on this?” And you might find something interesting. So, that’s the looking at the future is, what are the different trends? And one obviously massive set of trends that I look at is technology, but by the way, it’s also markets. It’s investment and capital. It’s competition and different workforces. It’s changes in regulatory law or social considerations. And you look at all of those things and you go, “Okay, what might happen? And if that might happen, how should that change what I’m doing now?”
ADI IGNATIUS: So, I want to talk a little bit about the man versus machine. And as you just said, there are pessimists who are concerned. You’ve actually argued that instead of this man versus machine, that technology makes us all more human. Can you explain how that works?
REID HOFFMAN: So, it goes all the way back to why do we have … the centers of human progress have been the creation of villages and cities, and that’s a result of technology, agriculture. We go, “Okay, this is technology. This is technology. We’re speaking through technology.” All of these things change the pattern of how we live, how we do business, how we create products and services, how we market them, deliver them, how we do financial analysis, how media communicate, all of it. And so, opposed to thinking of ourselves as homo sapiens because we go, “Oh, what’s really important is how we think.” Of course, it’s very important, but by the way, the pattern of sapiens and how we think is technology. So, for example, once we get to the written word, once we get to the printing press, once we get to the Internet, once we get to podcasting, all of these things are, we evolve through technology. And that’s true at the macro level, the history of human societies. And it’s true at the micro level. For example, if you say, “Well, okay, I’m a writer, but I don’t use word processors.” Well, it’s possible. It’s very hard. I’m a graphics designer. I don’t use Adobe or Figma. Very hard. And so it’s the question of these tools are how we progress, and we learn, and the AI tools are going to take every single tool that you use in a professional context, even email, or meetings, or other things, and is going to bring them to the next level, massive new transformational features. And that will be happening over a small n number of years for that. And so, what does that mean for you? What does it mean for your organization, etc.? And that’s the person plus machine. Now, what happens is people say, “Well, shoot, if I make my marketing people 10x more productive, am I going to, for example, lay off 90 percent of my marketing people?” Well, look, sales, marketing, those are competitive functions between businesses. The whole point is to be a lot stronger or relative to your competitors as you’re doing things. I don’t think you’re going to lay off any marketing people. I actually don’t even think you’re going to lay off any salespeople. What you’re going to do is you’re going to say, “I’m going to refactor the throwaway to my department for that.” Some of that may be trading some people who cannot use the new tools. Yes, that’s part of the job transition. But some of it is upskilling them, and some of it’s hiring some new people to do it, and it’s using these new tools in that competitive game that is sales and marketing. And when you look at the whole swath of an organization, you say, “Well, I could use a lot more engineering.” You say, “Well, what about legal and finance?” It’s like, “Well, actually, in fact, we do the amount of legal and finance that we can afford in our organizations.” Previously, when spreadsheets, VisiCalc was created, people said, “Oh, accountants, all these people were doing double entry accounting and calculators and books, they’re all going to go away.” And yes, that function went away. Accounting didn’t go away. Accountants now do much more detailed analyses about risk factors, and scenario planning, and revenue attribution, and financial modeling, and all that. And that same thing is going to happen when you get to legal, when you get to finance, and so forth, as part of how AI works. So, it’s that exact human-plus machine, and that’s why I call us homo techne. And that’s the natural pattern of it. It doesn’t mean that there aren’t lots of transition issues, and those transition issues are real. One of the nifty things about AI is that AI can also help with the transition issues. So, you say unlike, for example, I go to the account and say, “Here’s this new thing, VisiCalc,” and it’s like, “Oh, shit. I’ve got to learn this.” And some people are like, “Well, here’s this new thing, AI. And by the way, the AI is going to help you learn it too.”
ADI IGNATIUS: All right, so let’s spin forward then. So, as you think about the future and 10, 15 years out, do man and machine merge? I mean, how do you imagine the delivery of all this and the merging of man and machine? What does it look like, do you think, in 10, 15 years?
REID HOFFMAN: In 10, 15 years, a guaranteed way to look a little foolish because the future is almost always sooner and stranger than you think. And so, obviously, some science fiction technologists will go, “Wow, we’ll have the equivalent of neural implants and other kinds of things.” And some will say, “Hey, it’s now like an earbud or something as a way of doing this.” I tend to think that when we see this, we tend to be ant. So, we’ll still have a phone. We’ll still the computer or a display interface, computer, and so forth. But it certainly would … I think part of what generative AI is going to do is the interface to all this stuff becomes much more natural language. We’ve had to learn when we interacted with these devices because they’re very brittle in terms of they need an exact semantics. And part of what we built is like a windowing environment and so forth to make the semantics much more naturally translatable, like I’m pointing at this, and I want to turn on the mic here, and I want to readjust the sound there, and other kinds of things as ways of doing it. And now, with AI, language is not so brittle. We’re going to use language a lot more. And so, I tend to think of when we’re doing precision, we’ll be typing. And when we’re like, “Oh, I want to write a memo about AI’s impact within the manufacturing industry and how action-packed the information ecosystems will actually, in fact, have the following kinds of efficiency and da-da-da.” You’re just throwing it all out there, and it then starts acting on it. And so I think a whole set of the devices, the complete interface all the way down, is going to change on that. And so if you ask me to guess what the newest thing might be is a generalization out of the equivalent of earbuds where you are using language fast and a lot more efficient, quick on what you’re doing because, for example, almost all of us can speak a lot faster than we can type. And part of the reason why we type is because it gives us a precision different metaphor. But when the thing is interpreting you and getting your interpretation right, then it doesn’t really matter if your sentences are not fully grammatically correct while you’re speaking. You literally will just start going … for example, like I was doing that memo on manufacturing, you go, “Memo: manufacturing, AI, information supply chains, increased efficiencies, da, da, da, da, da.” And it goes, “Oh, you’re looking for something like this, da, da, da, da, da.” And then, it’ll start. And our thinking and patterning becomes much more like video game playing in terms of the way we operate than the current discourse of I sit in a room, and I think for a couple of hours as we do it. I’m not saying we won’t be doing that, but all of a sudden, we have this new way of using the tools. And that new front-end tool, I think, will be primarily language-based. Now, obviously, some people are working on glasses because you see the whole world. And when I look at the screen and look at Adi, it’ll go, “Oh, ask Adi this question,” or something. And, obviously, we’ve seen Minority Report, this movie, and other kinds of things as ways where these come out, but audio first, I think.
ADI IGNATIUS: So, let’s jump back to 2024 for a second. So, you’ve had fun, I guess I would say, with the fake Reid Hoffman, using AI to create an animated likeness. So, I’m interested. What have you learned from this experience about the ability to do more with your brand and likeness, and what have you learned about the ability of others to use the technology to hijack your brand and likeness?
REID HOFFMAN: Well, we always have some serious issues with deepfakes and so forth. We’re in political season. There’s obviously the set of things that are happening there. I think that one has to be careful about that kind of stuff. Now, part of the reason I did it is I wanted to show, even in this what is obviously very worrisome technology because deepfakes are worrisome across everything from political elections to what’s going on with the harassment of teens and other kinds of things, is you go, “Okay, what’s the set of things that can also be very positive?” And like, for example, one of the things in terms of communication, why are people here? They want to get a sense, not just of what I’m saying, but who I am and all that kind of thing. I, unfortunately, only have the ability to speak one language with any real coherence. And yet, when I gave the honorary doctorate speech in Perugia, I had Read AI give it in Italian, in Hindi, in Chinese. And so you have that kind of human connection and that capability of not just translating it in those languages in AI but delivering it that way. You get a broader brush of … I can only select very few events like this one that I can actually speak at. I’d say I probably turn down 99 for every one that I accept, maybe even more. And yet, now that I have the other thing, I say, “Oh, that’s really … that’s a great young student group at the University College London. I can allow the Read AI.” Or, for example, that Eric Brynjolfsson teaching that business school class at Stanford. I’m out of town. I got a career commitment, but Read AI can come and do that and help and participate. And so those are all the ranges of the positive things. I’m not yet the recipient of a negative use of a deepfake. So, is it a digital twin or something else with reading? But obviously, those things will happen, and we need to navigate them as a society.
ADI IGNATIUS: Yeah. So, I want to go to a couple of audience questions. This is from Jaime, co-founder at Latin Leap. So, the question is, given your deep involvement in both AI and the startup ecosystem, how do you envision generative AI transforming the entrepreneurial landscape?
REID HOFFMAN: Oh, it’s massive. So, on one hand, you get to the professional thing of every individual should be trying to use GPT-4 and all the others, a number of the others, as a features or attributes. Entrepreneurs tend to be adoption of new technology. So, it’s like, well, doing my research, doing my PowerPoint, doing my marketing materials, doing my sales materials, figuring out how to recruit, figuring out which venture capitalists I should talk to. So, on a micro level, it’ll just be massive across all these fronts. Then on, okay, well, what kinds of products and services? Because part of what happens in entrepreneurship, and this is to your earlier question, Adi, of anticipating the changes in future. Part of what opens up entrepreneurship is new markets, new products, new services. And you say, “Well, what’s going to happen because of this that the other incumbents or maybe other startups haven’t seen yet that I can see and I can go operate on?” And classically, new platforms open up room for new massive technology companies like the Internet gets to Google and Amazon, and then Web2.0 gets to Meta. And obviously, Meta gets a lot bigger with its successfully transitioning to mobile. And then, of course, you have Apple reinventing itself with the iPhone. Each of these kind of things, these technological platforms, that creates entirely new ecosystem in areas. And the startups, Google, Amazon, Meta, do that quite well. AI is going to be doing the same thing. Now, it doesn’t mean that creating your own frontier model of the supersized scale is the right thing. That actually, I think, will very quickly not be a zone of startups but only a zone of the hyperscalers, or most likely 99-plus percent. But on the other hand, a whole bunch of startups that I’m investing in that are transforming productivity, transforming medicine, transforming applications within the enterprise, and all of those things that my partners and I at Greylock are looking at doing, those create massive new startup entrepreneurial openings and activities. And then there’s this question around, for example, the surfaces of existing products massively change. Maybe companies will adopt it the right way, or maybe they won’t. And when they don’t, that’s frequently when a challenger can challenge an incumbent and make a new product area, whether it’s a CRM product or something else. And so I think that’s all in the entrepreneurship arena.
ADI IGNATIUS: So, here’s another question from the audience. This is from Sunchit, who identifies as a student, who asks since AI is made by humans, do you think that AI ultimately will be limited to the human capacity? In other words, human folly becomes AI folly, or do you think it could transcend these limitations?
REID HOFFMAN: Well, we definitely already have pretty good examples about how technology made by humans adds in superpowers that are beyond what are human superpowers. And it isn’t just in the physical arena. Try driving in a car versus running, but also mental, like computers and their ilk add a huge amount of mathematical capabilities to what otherwise would be mistakes. Now, you say, “Well, look. The deeper version of the question is, are there tendencies for human beings, for example, to have confirmation bias and decision?” And so, does that lead you to the kind of thing that gets replicated within the devices? And the answer is some and some. So, obviously, there will be blind spots that we won’t see our way around that we’ll have to work on in order to do just like we do with collective human beings. And that’s part of the reason why we do a lot of work joint because it helps us, not just the number of hands, but the collective cognitive capabilities and the diversity of those cognitive viewpoints bring in to solve problems. But also, there will be a bunch of areas where that will help. So, for example, if you say, “Well, I have a confirmation bias, but now I have an agent that’s looking at the information with me and goes, ‘Well, you do know there’s some really good things that contradict this confirmation bias.’” They go, “Oh. Well, tell me about them.” So, I think it’s, generally speaking, a lot better, and not just it’s only replicating our biases. But it doesn’t mean that we can’t just rely upon it, and that will be an ongoing work.
ADI IGNATIUS: Yep. So, we probably only have time for one more question. This goes fast. But this is from Lena, an IT business analyst. Lena is picking up on the techno-humanist ideas of yours. So, the question is, what does a successful techno-humanist society look like … again, it’s out in the future … in the next 10 or 20 years? What are key milestones that one should look at along the journey?
REID HOFFMAN: What I would say is how do we have technology amplify what do we think are the key characteristics of human existence? And one way to look at it in terms of the virtues is when we look at the inverse of what we think of as inhuman. So, we go not just intelligent, or not just smart, or not just quick, but also wise, compassionate, etc. And how can all of the various technological elements and how we navigate individually, but also navigating groups and navigating society combine to making us better? And obviously, when we look at a bunch of mostly video media, everything from Black Mirror to those of things, they all come down to man versus machine, machine ruining things. But actually, in fact, when you look at it, what we’ve done over time, there’s a transition period. And so we had to get through the Industrial Revolution. We had to integrate the printing press and all the rest, but we get through it, we suddenly get to a much more elevated human circumstance where we’re better educated. We have better perspective. We have room and chance for dialogue on different points. And I think that’s the continuation of the thread, and I think that’s what we’ll want to see with … we’re going to have millions of AI agents. And how do those millions of AI agents help us with that?
ADI IGNATIUS: So, I’m actually going to ask one more question. This is from Kristin, who is CEO at The Pad Climbing, and it comes at all of this from a different angle. Kristin says, “AI could easily allow us to ask for best practices like a mentor, but it will come up with basic solutions.” And some of us have experienced that generative AI is incredible, it’s fast, but often, the text is generic. So, the question is, how do we make sure we’re infusing the creativity that’s needed to build the network so it isn’t predictable and generic but is actually something creative and remarkable?
REID HOFFMAN: Well, one of the skills … there’s a whole long set of answers to this, but even the current AI tools is pay attention to the skill of prompting it the right way. It’s like if you come and ask someone like me and say, “Well, what are the things that create an enduring business?” You go, “Well, it’s a really good set of products and services that have good operating margins and good competitive modes.” And you’re like, “Yeah, yeah, blah, blah, blah, blah, basic.” But if you say, “Well, what are places where adding intelligence and artificial intelligence transforms all elements of how a normal business operates?” Then you get a better answer, the same thing as in prompting these AIs. And so, for example, I very rarely just go like, “Write me an investment memo in AI.” I will go, “Write me an investment memo on this. Highlight the following possibilities or concerns. Make sure that you address the questions around capital efficiency in model deployment and where models are going, etc.” And then you get something much better. And so, for more pizzazz, frequently, if you prompt them to say, “Give me three very different answers to this question,” and then, all of a sudden, you’re less in the vanilla mainstream. And that’s just the beginning of looking at how we use these tools, what the prompts are, and all the rest.
ALISON BEARD: That was Reid Hoffman, the co-founder of LinkedIn and Inflection AI, in conversation with HBR editor-in-chief Adi Ignatius at our Future of Business virtual conference. I hope you listen to all of our Future of Business series and all of the episodes we have on the HBR IdeaCast about leadership, strategy, and the future of work. Find us at HBR.org/podcasts or search HBR on Apple Podcasts, Spotify, or wherever you listen. And if you don’t already subscribe to HBR, please do. It’s the best way to support our show. Go to HBR.org/subscribe to learn more. Thanks to our team, senior producers Anne Saini and Mary Dooe, associate producer Hannah Bates, audio product manager Ian Fox, and senior production specialist Rob Eckhardt. And thanks to you for listening to the HBR IdeaCast. I’m Alison Beard.
