Reid Hoffman: Should we trust the tech elite?

0:00This BBC podcast is supported by ads outside the UK. If journalism is the first draft of history, what happens if that draft is flawed? In 1999, four Russian apartment buildings were bombed, hundreds killed. But even now, we still don't know for sure who did it. It's a mystery that sparked chilling theories. I'm Helena Merriman, and in a new BBC series, I'm talking to the reporters who first covered this story. What did they miss the first time? The History Bureau. Putin and the apartment bombs. Listen on BBC.com or wherever you get your podcasts. Welcome to Hard Talk from the BBC World Service with me, Stephen Sacker.

0:46My guest today is a member of the American Technology Billionaires Club, the so-called Tech Bros, whose businesses have transformed the way we live and work in a single generation. Reid Hoffman made his first fortune as a founding board member of the digital payment system PayPal, where he worked alongside two other tech pioneers, Peter Thiel and Elon Musk. Back then, they were close. Now, not so much. Hoffman went on to found LinkedIn, the social media platform dedicated to professional networking, now used by hundreds of millions around the world in pursuit of career advancement. In recent years, Hoffman has been a major investor in artificial intelligence, and he currently sits on the board of Microsoft, Inflection AI, and a number of other AI ventures.

1:35For many, inside the tech industry and in the world beyond, the rapid advance of AI capabilities, far beyond the large language models capable of generating sophisticated text, raises troubling questions. As machine intelligence becomes ever more potent and autonomous, what happens to human agency and value? Are we unleashing a power we cannot or will not control? Well, Reid Hoffman joins me now. Welcome to Hard Talk. It's great to be here. Well, it's great to have you here. I wonder if you would acknowledge that right now there is something of a crisis of public confidence in technology as a force for good.

2:15I think there is, although I think there shouldn't be. And it isn't because I'm just, you know, I spent my life in the tech industry and I build companies, LinkedIn, as you mentioned, and others and invest in them. It's because when you look at where are causes for hope in terms of how is the next decade and the decade after that better, a lot of it comes from technology advances. And we want to reap that benefit for society and humanity. But we have to trust you when you say that. And perhaps trust has been eroded in recent years, maybe one could argue the last decade, as we have seen that some of the big promises made at the dawning of the age of the Internet, digital technology, some of those big promises haven't been kept.

2:57Well, I think there's a lot of promises that have been kept. I think there's some, we're still in some birthing issues and some transition issues. You know, I myself have kind of argued that, you know, social networks should look more like LinkedIn than the other ones. And, you know, the kind of question about how you embrace how we collectively learn as they, how they work. But the fact is they look as they look. And to many people, they are corrosive and they are damaging young minds. And one could go further. One could say that the very basic premise at the beginning of the Internet age, that it would somehow empower individuals, it would democratize, it would break down hierarchies and empower individuals.

3:32In many ways, that isn't happening. In fact, one could say it's empowering states, authoritarians, people who want to manipulate the individual. Well, I think it's empowering both. So I don't think it's not empowering individuals, But I think it's also have to navigate how does it work within society? How does it work within authoritarian states? How does it work within the question of how democracies operate? But by the way, part of not to lose track of the reason I was saying the Internet has maintained a lot of its promises. Think about like Wikipedia and the access to all this free information about tons and tons of subjects, about life, whether it's medicine, learning, you know, economic opportunities like LinkedIn.

4:10And that's part of the reason why it's like, no, no, no, there are some challenges we're still working through. Absolutely. But let's not also forget about a lot of the good things. No, I understand that. But then again, you would say that, wouldn't you? I referred to you as part of the quote-unquote tech bro elite at the very beginning of this. And that's perhaps another way in which the public is losing confidence in the messaging of people like you. Because the truth is, the greatest beneficiaries of this digital age of ours are the very few ultra-wealthy entrepreneurs and brilliant minds who have made unimaginable fortunes, have become oligarchs of the tech era.

4:49And many others don't appear to be benefiting at all. Well, I think it's the nature of human society that some people disproportionately benefit. And it's not just now with the tech industry. This is a disproportionate like of which we've never seen before. I mean, to quote Ted Chiang, and you quote him in your book, Super Agency, to your credit, it doesn't necessarily reflect well on your industry, but Ted Chiang says that basically what we have seen is that the internet and all that goes with it has served to increase the wealth of the top 1 % of the top 1 % instead of raising the standard of living for US citizens as a whole.

5:27Well, I think, unfortunately, in that regard, TED is not 100 % correct. I think that there is a question about where, you know, look, entrepreneurs get wealthy. That's been true, not just with the internet industry. That's been true for energy, railroads. And actually, the people who say this is the most ever, I usually say, well, what's your historical analysis? And usually they don't have it, right? Because it's kind of, it's more of a rhetorical device than it is actually a substantiated piece of argument. Now, that being said, when you say, well, how has it benefited your everyday citizen?

5:58And it's like, well, your everyday citizen is using the Internet every day. They're using it to discover information. They're using Google Maps. They'll walk around the city and discover stuff. They're using it to determine which products to buy or not. There's enormous amounts of benefit there. That doesn't mean that it's perfect. There isn't things to fix. But it's only been beneficial to, you know, call it 30 or 40 people who started companies is simply incorrect. We are going to talk in detail about artificial intelligence. One of the great fears about the development, evolution of AI is that one of the first things it's going to do is rob people of their jobs.

6:33You seem to be the tech optimist who says that's a complete misunderstanding of the way AI is going to work. But the fact is Elon Musk and many others who have been great advocates of AI say that, yes, 10 years from now, many, if not most of the jobs that currently exist will be done by machines. So when you look at this, I actually think what you'll find is that people doing jobs will be replaced by people with AI doing jobs. And part of what I think we want to be doing in society is we want to be helping those people who are doing jobs be the people who are replacing the person doing the job with themselves using AI.

7:07And one of the benefits of AI is that AI can help you learn how to use AI, can help you do the next set of tasks in the job. So now I don't want to undercut that transitions are difficult. Part of what I say in superagency is artificial intelligence is the cognitive industrial revolution. And I want that both to indicate the power of it in terms of the benefit to the economic prosperity of not only the society, but your children, the generations to come. If you don't embrace the industrial revolution, you fall behind in global economy. On the other hand, the transitions are never easy. And the industrial revolution transition was hard.

7:43And part of the reason I say this is let's try to make that transition better. I'm not saying it's going to be easy, but better than we've done earlier other transitions. It just seems to me something interesting is happening in AI, which is in some ways analogous to what happened with earlier sort of phases of the internet revolution. That is, a lot of idealism is being overtaken by a commercial profit motive. If one looks at OpenAI, an organization that you were involved, I believe, in as a supporter and investor from the beginning. It began with this sort of mission for humanity, and it was committed to being not-for-profit.

8:21Well, all that's gone out of the window. I mean, now OpenAI, as I understand it, is absolutely about maximizing profit, and it's in a huge sort of hookup with Microsoft, where Microsoft, I think, gets 49 % of its profit. So, you know, things are changing even as we speak within AI, and it is becoming much more an openly commercial evolution. Well, part of OpenAI's journey was it's still on the mission of how do you build AGI for the benefit of humanity? It needs capital in order to do that. The philanthropic capital market was insufficient, so converted to what we call in the US a public benefit corporation.

8:57It's like a corporation by limited guarantee or something on this side of the pond. And that allows its board of directors and its executive staff to treat its mission as more important than its profit and how it's executing. So it continues to focus on its mission, but the way to execute its mission, like many missions, is through actually what it needs is the capital that the for-profit markets bring to bear. Well, that seems a complicated way of saying that big tech in the name of Microsoft is now a hugely influential player in what OpenAI chooses to do, and they want profit. They want a return on their investment.

9:33Well, I think the deal works fine, but Microsoft is not on the board of OpenAI. OpenAI has its own independent board. Like I said, it's a public benefit corp. So the instructions, I've been on a number of boards, for the board to follow is following the mission. This is you talking about AI and you're describing it as, quote, a desire to create new ways for people to attain more agency in their lives. And of course, you called your book super agency. But to many, AI is absolutely not about human agency. It's about machine agency. It is the opposite of human centric. So in these transitions, you always start with a transformation of agency, which feels like a loss of agency, not just because it's like, well, what if I choose not to engage in this transformation?

10:24Why is a transformation going to happen anyway? And the answer is it's going to happen anyway because technology will be built around the world. Companies will be competing with each other. Industries will be competing with each other. And just like any productivity metric within a kind of the free society and the free market, it goes that direction. Now, on the other side, just as when we got to the other side of the Industrial Revolution, you have a mammoth increase in agency. Sure, the person who was driving the horse and buggy carriage or the person who was the independent weaver in their own house, the agency of those jobs has reduced, is no longer there.

10:58But the new ability to, for example, we have a large middle class that sustains a democracy, sustains an educational system, all come about because of the Industrial Revolution. And that's what I'm saying is, as you get through this transition, you get a much stronger agency. But if you want human agency to be paramount, surely you would want human regulation and oversight of the rollout and the ever more complex evolution of artificial intelligence. And yet you seem to be an enemy of regulation. Well, I may be focused on regulation for knowing that what you're creating when you're innovating is actually going to be so important.

11:37Because this is one of the problems that generally Europe runs into, is to say we can pre-plan innovation. We can have a heavy regulatory regime that dictates where software should go. And it's one of the reasons why Europe generally doesn't have much in the way of interesting software companies building platforms, etc. You actually have to allow the development and the experimentation and failures as you're going in order to do that. Well, hang on, but you don't have to sit by and watch as harms deepen as a result of technological development. And we have seen harms from the very beginning of this discussion.

12:12We've discussed some of the harms that have come with the digital age. And Europe's position is, no, we need to be proactive in ensuring that the public is protected from these harms. Well, the question is, you will have some harms if you allow innovation to happen at all. If you say we're going to slow down innovation so much that we allow almost no harms, then your innovation will be much less than your other countries, your other industries. Sure, but these aren't just any old harms. We're talking about fundamental invasions of privacy. We're talking about the manipulation of young minds. We're talking about the spread of disinformation, which can have massive political ramifications and does in different parts of the world.

12:54These are harms that societies can't afford. Well, but the question is, how do you navigate doing them? If you say we must simply have zero of them, then the other people who venture into it, take the risks, take some of the pain in going through it, then build those tech companies. And then you just have to follow those tech companies. And I, for one, would like those tech companies to also be built here and also be built in Europe. It's a strange sort of argument because you seem to be arguing that in a funny sort of way, the evolution of tech suits non-democratic authoritarian regimes better than it suits democracies.

13:30And that democracies can only thrive in technology if they, in a sense, control and reduce some of their democratic instincts. It's not democratic instincts. And by the way, authoritarian states also frequently exert strong regulatory control to some limitation. It's a allow innovation and allow some risk and then fix it as you go. So one of the things I talk about in super agency is iterative deployment. And I use chat GPT as an example. So people go, oh, my God, I don't know about AI. But you can start using it and you can see how it can benefit you. And that's actually part of how we bring the public into the loop.

14:07As opposed to having, for example, a government regulator who's sitting behind a desk who says, well, I think I know better for how industry should go, how technology should be developed. I think I can channel, what, tens of millions of citizens. It's let's engage people and learn from that and then iterate as we go. And that doesn't mean no regulation. So, for example, the car industry, if you said, before we deploy a car, you must guarantee me that there is zero harm. You'll never deploy a car. You deploy the cars. And that, again, doesn't mean no regulation. Because, for example, cars are deployed.

14:35No one wants seatbelts. The consumers don't want seatbelts. And so then society has to step in and say, no, no, no, seatbelts are important. And you add that regulation in. But you add it in as you're developing and building, not as you try to do it before. Do you believe, if you look at what's happening right now, that the Chinese are not only catching up with where the United States is, but are actually in the process of overtaking? I don't think they're in the process of overtaking, but I think they are fully in it. And this is actually one of the problems when you get to how should we adjudicate what is the future of copyright law.

15:09Because you could say, well, OK, if you're training a model within the Western an ecosystem, we should modify, which I think would be changing copyright law to disable, essentially, AI training. The Chinese won't follow that, and they will produce open models like DeepSeek. So once again, you'll have a sense of, okay, well, that won't answer the concerns and needs of the creative community. The Chinese have announced that there've been major developments with something that they're calling Manus AI, which they're saying is a much more autonomous form of artificial general intelligence, which seems to cross over to something new.

15:48And I guess the key word is much more autonomous. Are we on the cusp of something really important here, which again, it seems the Chinese may be moving on, certainly in public, faster than companies in the US? The labs that I'm part of don't look at Manus as kind of cutting edge work. We actually already have all of that work. We may be slower in deploying it because we have all kinds of safety things to make sure that individuals and societies are... We do as much as possible to avoid harm. But the notion of having some autonomy of an agent... And by the way, that's like, go book a plane ticket for you.

16:30That's, take actions on the internet for you. That's technology that's also being built by all of the major tech companies in California. You say we are so careful not to do harms. The truth is there are many people in this space who are deeply worried about how fast it's going and how far it's going. Demis Hassabis, I'm sure you know, won a Nobel laureate for his work at the Origins of AI. He now says it's time for an AI sort of global supervisory agency like, a little bit like the IAEA. That's the International Atomic Energy Authority, which controls or tries to control non-proliferation of nuclear weapons and weapons of mass destruction.

17:11The analogy is quite clear there, that there is a potential danger with the rollout of AI, which could be akin in terms of threat to a potential weapon of mass destruction. And he says it needs to be controlled. Look, I think there's levels of AI that could certainly get there. And I think that the notion of, and the principal thing that Demis and I talk about this is how likely is it we could get global cooperation? If I was a big believer in the global cooperation, which, you know, was roughly speaking, you know, think about global cooperation and climate change. You can see the impact much clearer.

17:49It's not the vague possibilities of some possible future branches around AI. You see the temperature getting higher, and yet you don't have global cooperation. So I tend to be a little bit. You're not saying it's impossible. I would love it to be possible. I have not yet seen it. If it's not possible, then it comes back to where we began, which is about the level of trust we can have in the leaders, particularly in the Western world of this particular technological field. Let's bring it back to you. And you have made it clear that you act for humanity. you've fallen out with tech elite members whom you used to work with like Elon Musk who you now describe as deeply irresponsible.

18:31I think you called Musk nutty at one point. Yes. You have got into a major spat with the guy. Do you believe that people like Elon Musk can be trusted? Well I think he's been trusted with space entrepreneurship. He can be trusted with car entrepreneurship. I think he is behaving irresponsibly with government. I think dealing with inefficiency in government is a very good thing, but I actually think that one can do it in a less cruel and a more lawful manner than what he's doing. And I don't think that running governments is like running corporations. I like corporations. I've started them, I've invested in them, and so forth.

19:08And I think they are an important force for change within society in a positive way. On the other hand, that's not how government should work. And so I think as it crosses over into that, I have much less trust. But something interesting is happening in the United States, which is a democracy, but where more and more of the tech elite appear for one reason or another to be lining up with Donald Trump with the power in the White House. And in the case of Musk, he's with X become the key propaganda arm of the Trump White House. Why is there, this lineup of power and technology right now in America?

19:48And how dangerous could that be? Because you're not part of this. In fact, you swim against that tide. So I think there's two different challenges here. One is, I do think it's the responsibility of any industry, including the tech industry, to be responsive to democratically elected governments. The administration is democratically elected. And so every industry should be responsible. So the fact that a bunch of tech leaders are being responsive is, I think, democracy as it functions. Now, there's a separate question about how the government itself is operating. And the government operating, I would like it to actually, in fact, be more consultative to the kind of set of considerations that go into a government.

20:28So, for example, one of the ways that a corporation and a government's different is, I can take a risk with a corporation and the corporation can blow up. And that's a problem for employees and shareholders and so forth, but isn't necessarily a problem for society. If government blows up, that's a problem for all of society. But you say you really believe in democracy. It's important that the richest titans in the world believe and support democracy. But you've been accused of actually manipulating the public using disinformation techniques yourself. I mean, you've accused Musk of it, but you do it yourself.

20:58You did it in a famous Senate race in the United States in 2018, where it turns out your money was used to manipulate voters, give them disinformation about a Republican candidate, Roy Moore? So first, I challenge your characterization that I did it, right? Second... It's your money that backs the campaign, which turned out to be using disinformation techniques. I gave money to an organization to do something that was more study what's going on with social networks. I learned about their claims that they used it for a false flag operation in the same way you did by reading the New York Times and never actually saw that.

21:36When I saw that, I apologized for my money being misused. And I said, I will now, which I have done, put more constrictions on, you must give me reports about how you're using the money. And that's why I challenge your contention that I had anything to do with it other than accidentally funding an organization that claimed it did bad things where it's still unclear what exactly it did. I guess I mention it because as we end this interview, I just want to come back to the broad question of trust and whether the public can have confidence that people like you, like Elon Musk, like Bezos, like Zuckerberg, who are so influential right now, truly will, over the next five to ten years, as you plow more money into the next stage of tech advance, which really clearly is AI, whether the public can be confident that you will do it in a way that is responsible and that actually is for the public good.

22:27So I think that the way we do it is the way, the good way is the way we deploy in the book, right, which is what we call it is the consent of the governed, which is how do we get lots of people involved in experiencing the products? How do we have transparent discussion about what we're doing and what we're building? How do we accept critique about, well, actually, in fact, you're doing X and these people would rather you didn't do X? And by the way, anything that affects millions of people, you'll always have some people who disagree with it. That's the nature of society. That's everything in society when you get to millions of people.

22:59And so you're trying to navigate it as best you can. And I think that there are a number of entrepreneurs, I hope I am one of them, who are doing that with a view towards what could make society a lot better. And if you said, okay, what would I argue as why one should potentially trust me? I'd say, look at LinkedIn. And if you don't trust LinkedIn, that's fine. There are probably amongst a billion people who've registered for LinkedIn, there are probably some people that do it. That's totally fine. I think many people would say, hey, this is a really valuable service for me. And this is something that's really made a good difference in my life.

23:36Reid Hoffman, we do have to end there. But I thank you very much for joining me on Hard Talk. Thank you. If journalism is the first draft of history, what happens if that draft is flawed? In 1999, four Russian apartment buildings were bombed, hundreds killed. But even now, we still don't know for sure who did it. It's a mystery that sparked chilling theories. I'm Helena Merriman, and in a new BBC series, I'm talking to the reporters who first covered this story. What did they miss the first time? The History Bureau, Putin and the apartment bombs. Listen on bbc.com or wherever you get your podcasts.