145. Hoffman: Europe Is Regulating Itself Out Of The AI Race

0:00Now, I'm delighted to tell you that Monzo Business are now a proud partner of The Rest is Money. We're teaming up for the next six months. You'll be hearing why they're a great option for your business banking and how they're supporting small businesses up and down the UK. Now, since I am a Monzo business customer already, I can tell you all about it. They've got three different plans to choose from depending on the size of your business. So, for example, with two of us in mind, me and my business partner in our latest venture, we've opted for the pro version, which comes with a load of tools to help us simplify all of that financial admin.

0:37There's also light for sole traders and team for those slightly larger businesses that are scaling up. So why not join me and over 500 ,000 businesses already banking with Monzo Business and sign up for an account today. Only sole traders or limited company directors in the UK can apply. Invoicing is available with Monzo Business Pro for£9 a month and team for£25 a month. T's and C's apply.

1:14Hello and welcome to The Rest is Money with me, Steph McGovern. And with me, Robert Peston. So we have got a brilliant guest for you today, Reid Hoffman, co-founder of LinkedIn. He is one of the giants in the tech world who was part of, you know, early investor in companies like Facebook, OpenAI, Airbnb, also part of what's known as the PayPal mafia. This is this group of people who started PayPal and then have gone on to found some of the biggest tech companies that we love talking about on this show. Or, you know, we're talking in this gang is Elon Musk, Sam Altman, Peter Thiel as well. They're the kind of mega giants of industry, of the world at the moment.

1:56The superpowers, aren't they? They seem to be running everything. So he's a hell of a person to speak to. He's got a book out, Super Agency, which is all about trying to explain the positives of AI and what we can learn from it, how it's going to change the world, which is kind of right up our street as well. Yeah, that's right. Right. And, you know, as somebody in his early years work closely with Musk, obviously his take on what Elon Musk is doing to reengineer in this brutal way the entire American government, that's going to be fascinating to talk to him about. But then there is just this central question as Europe is forced to re-evaluate everything from defence to, you know, how it generates prosperity because of the way that Trump has effectively distanced America from Europe.

2:50This issue of how Europe engages with the AI industrial revolution is more important than ever. So this feels like a pretty important conversation. Yeah, certainly does. Here's our interview with Reid Hoffman. Reid, thank you so much for joining us. We have got so much to talk to you about. Obviously, we want to talk to you about AI, your book and everything else. Before we get into that, we've talked a lot on this podcast. Everyone's talking at the moment about what's happening in the US. I mean, it's scaring the life out of everyone. What's your take on it? Well, unfortunately, I think fear and concern and anxiety are very good rational responses.

3:35I mean, obviously, for example, the situation in Ukraine where Russia as an aggressor invaded Ukraine and has had a, to borrow a term from the crypto industry, a rug pull on it. And it's not only concerning, obviously, for the many thousands of Ukrainians who have lost their lives in this invasion, but also for Europe. You've got concerns around approaching a government efficiency, which is a really good thing. but doing it in a way that you're incompetently firing nuclear inspectors and Ebola folks and all the rest as part of doing this. And that's all deeply, deeply concerning. And so fear, unfortunately, here is a rational response.

4:19Can I just ask you, Reed, as somebody who has known Elon Musk pretty much your entire career, are you surprised by the extraordinary risks he's taking in the way that he is overseeing the reconstruction of the American government? Unfortunately, no. I mean, part of the thing that makes Elon a spectacular entrepreneur is he's willing to take risks of blowing the whole thing up, which when you're doing a company, it's one of the many reasons why company governance and government governance are not the same thing. When you're doing a company and blowing the company up, most of the time, that's okay.

4:56Society gets along just fine with any company, particularly big banks and a few others, but any big company blowing up, that's fine. The government blowing up is obviously a disaster. It's something that's akin to the kinds of collapse you have during the French Revolution or anything else, a number of different things. And so that's part of the reason why approaching the government like you manage a company is unwise in the extreme. That's before you get to questions of lawfulness, cruelty and other things. But I'm not surprised that he basically, you know, that he thinks, hey, I've done this successfully with some really amazing companies.

5:35So it'll work here. So why not? Where do you think it'll end, though? Do you think Musk and Trump are going to get along in all of this as this happens? Or do you think there's going to be a spectacular fallout there? Well, I think that this is actually, in fact, from a Trump perspective, it's great because Elon will go apply a wrecking ball. If it works, then Trump will claim credit. And if it doesn't work, he'll fire Elon. But firing Elon, surely that's not going to play well for him. Well, I mean, there's different kinds of firing, right? I mean, there's a, oh, you're not working on Doge anymore.

6:04You're working on our space program. As a sideways move. Because just to be clear, one of the things that we've talked a little bit about is Elon Musk's potential if he turns on Trump to do Trump harm is pretty spectacular, isn't it? Yeah, I think that, you know, the likelihood that Elon turns on Trump is quite low. Among the positive attributes of man is, you know, he's very dedicated to his particular missions, you know, getting people onto Mars. You know, I don't think it will be useful for those missions for him to turn on Trump. So I think that's actually, in fact, unlikely. I just want to ask on the sort of more fundamental point, you are criticizing Musk for the brutal way in which he is applying his business techniques to government, essentially closing something down to see if everybody can get along without it.

7:08You know, we've seen with the demolition of USAID already the harm that that's doing to vulnerable people in Africa, for example. But do you agree with Trump and Musk that the U.S. administration, the official side of the government is bloated and does need reform? Well, I think every American business person, including myself, agrees that one of the problems with governments everywhere, but especially the U.S. one, is that by not having the things we have in business, which is, you know, kind of pressures that cause us to have cost accountability. So we're always looking for how do we save money?

7:48You know, of course it's bloated. And I think that, but the question is, it'd be very easy. For example, a simple thing that if you were less cruel and more actually focused on the mission, you say, hey, we're going to take the first two months and we're going to ask memos for everyone as opposed to turning everything off. It's write me a justification. Of course, it's more work and you may miss some things because the justification sounds good even though there's inefficiency. But then you don't do things like fire people who are trying to prevent Ebola from spreading or fire people who are dealing with nuclear safety.

8:24Or, you know, for example, in the USAID case, the death toll is certainly going to be in the thousands, might be already, and may even get to the millions. And, you know, you didn't need to do that. You did that because you weren't willing to do the work for two to three months to actually sort it out before just turning off all the payments. It's interesting, your analysis of this in the sense of, you know, you can't just go in and treat a government, a public service sector like you can a company. Because there's loads of people who say that's what we need to do here. Like we need to run, for example, maybe the NHS.

8:57We need to run it as if it's a company rather than it's a public sector service. So can you just explain a little bit more about why? Because there'd be loads of people who think we need that. Well, there's definitely things. So corporations, companies are the best organizational form that we have found so far in the history of humanity for taking structured and scaling risks and working efficiently in that way. They're the most efficient organizations that we have thus find. So I'm very bullish on them and I invest them, create them, et cetera, a lot. And I think we should learn from them when we get to other sectors, whether it's government, healthcare systems, universities, a bunch of things.

9:39I think there's things to learn from them. The mistake is thinking they're the same. Yeah. And the one that I already kind of illustrated is, look, if a company goes boom and blows up and suddenly is basically just a smoking crater, it's a tragedy. Some people lost their jobs. Maybe there was parts of the ecosystem that were dependent on it. But it's very easily recoverable. If a government blows up, that's a terror for all of society. And it's one of the reasons why the governance and risk is different. And by the way, you then get some inefficiency because of that. People don't realize that you're buying the price, the insurance that you're paying on continued stability is some inefficiency.

10:21So you have to accept inefficiency. You have to accept some inefficiency. It's actually the nature of the beast and you're better off for it. It's like buying insurance. Yeah. It's like, oh, I'm going to do my company. I'm going to have my house, but I'm not going to buy insurance. It's like, whoa, if you don't have insurance on your house and it goes boom, that's a problem. Yeah, yeah. Can I just ask, Reid, as somebody who described Musk just now as probably one of the great entrepreneurs of all time, as I say, you have known him for decades. Why doesn't he get this? It's because to him, this is his hammer and everything looks like a nail.

11:01And, you know, he doesn't particularly consult with people. He doesn't particularly seek other expert advice. You know, he thinks that on almost everything he knows better than everyone else. you know I've I'm since these were private conversations I'm not going to talk out of school but I have heard a number of assertions where he goes I think x is true actually I'll do the one that's funny Elon once said when we were at a dinner everyone's smart I know knows we're living in a simulation right you know and and I was like um well what am I smart and he's like yes I'm like well then your statement's not true yeah yeah and then he what doesn't accept that yeah it's interesting because ai is your big thing as well isn't it and you know we've been talking about how ai could be the thing that revolutionizes everything for us but i keep saying about it being the next industrial revolution and you talk about it giving us superpowers if we harness it properly we can use that to change for good so do you think the politics aside that it's going to be ai that gets us through all this these troubles anyway and that gets us back on track?

12:06Politics can ruin everything, no matter what the technological solution. So I think one can never fully come out of how we govern ourselves, both as countries and as global orders, which I think the global order is also very important. But on the other hand, I think when you say, what's the things that we can do to make the best possible great things to elevate humanity over the next few decades? I, of course, think technology is that. And AI, it's one of the reasons why I wrote Super Agency. It's one of the reasons why I do the Possible podcast. This technology is probably the most significant technology that we have created thus far.

12:47Now, it builds upon it. You say, well, is it more significant than fire? Is it more significant than agriculture? Well, we wouldn't have gotten to AI if we hadn't done those. But it's incredibly significant because if you think about we can add intelligence and and intelligent assistance to everything that we do in the world. So for example, it's kind of a classic critique of AI is like, well, it's going to negatively affect climate change because we're going to have, be running these large data centers with a whole bunch of electricity. Well, if we're actually running these large data centers with a whole bunch of electricity on that, among the things we're going to be doing is saying how to figure out how to run the grid better.

13:25We actually already have proof points about how AI applied to electrical efficiency can, within the Google data center, really maximize DeepMind figured out how to do 15 to 35 % electricity savings there. So intelligence to everything helps everywhere. Yeah. And that's why it's so huge. And you call it, and essentially, if you harness this correctly, you can have a new state of super agency. What does that mean? So part of the reason that I chose super agency as a term is to say, you as a human being, is an individual human being are going to get superpowers with this. It already does. You could use it today for superpowers of things that most human beings just wouldn't necessarily know how to do.

14:06I mean, for example, here's a use case that most people don't do. Take a picture of something and just upload it to your favorite capable agent, whether it's chat GPT or anything else. The AI will figure out why you've uploaded a picture and then start answering what your thing is. Like you're looking at a microwave going, I don't know how to use this microwave. Take a picture goes oh you're going to try to use this microwave what are you trying to make and i'll tell you what buttons to push and what to do that's adding intelligence to everything and that's obviously a micro example which gets much more and more serious medical other kinds of things i've done it with a plant actually a plant and how did it work yeah it was really good because i've got a lemon tree that's dying and so i uploaded a picture to chat tbt and it told me that i needed to put coffee granules in it to bring it back to life and it totally worked Oh, yeah, there you go.

14:51Anyway, but that's literally the thing is you won't realize is you can add intelligence to everything. Yeah. And that's amazing. And including, for example, in a bunch of areas where people are critical of it. It's like climate change. No, no, this is going to help, you know, efficiency. And by the way, maybe in its most extreme cases, maybe it'll even help us with like getting better at inventing fusion, which is why technology is so important. Fusion plus geoengineering is the climate change solution. Anytime that we're not spending energy on that, we're actually not getting closer to climate change solution.

15:25I think both Steph and I share your optimism about the potential of AI to increase our living standards, increase the quality of our life if adopted and rolled out in the right kind of way. But there are also economic risks, as we've seen in other industrial revolutions. When you increase productivity in the way that AI does, some of that will translate relatively rapidly into higher living standards for those individuals. But some of it may lead to people losing their jobs. How do you see the role of government in a sense in maximizing welfare, in maximizing the benefits for people? So, and by the way, that just reminded me, I didn't fully answer the earlier question, but part of super agency is not just superpowers for individuals, but once a large number of people get the same superpowers at all at the same time, then we get superpowers as a society and it's part of super agency.

16:31And so, for example, as a car parallel, car helps, you know, I get a car and helps me be mobile and commute to work, you know, possibly live outside the city, et cetera. But then when my doctor gets a car and my kid is ill, the doctor can come to a home visit. And that's part of the super agency side. Now, for the government question, I think the industrial revolution is an extremely good parallel for what's happening with AI. I call it the cognitive industrial revolution. And I think that it's both from the amazing upside. We don't have anything of our current society. We don't have the middle class the way we have it.

17:05We don't have higher qualities of life, et cetera, without having gone through the Industrial Revolution. But the transition period was obviously very, very difficult. And I don't think there's any way to make this transition period like not also painful. I think the question is, is how do we learn both from the previous transition like industrial revolution or electricity or anything else? And how do we make this more human, more graceful, not zero pain? So not zero pain, Reid. And if you look back to the original industrial revolution, we had this long period of stagnating wages for hundreds and hundreds of thousands of people.

17:43what will be the pain that we are going to experience over the next few years? And then what would that mean, for example, to redesigning welfare systems, redesigning education systems? So look, I think the pain is going to be you add intelligent co-pilots to everything. That basically changes the vast majority of jobs. Some jobs will even actually just kind of go away. Like, for example, I think any job that we have, human beings trying to act like robots. So, for example, if you look at customer service jobs, customer service jobs are like I'm following a script on the computer. It's like, well, an AI can do that better.

18:20And so that's an instance of a job that will go away in 90%, 95 % as an instance. Other jobs will transition. Let's take, for example, graphic design and illustration. We already know that we have a variety of these tools. I don't think that means graphic designers will go away, but it's supposed to like, hey, I give a brief to somebody and I say, oh, I'll go away for two weeks and I'll come back and I'll show it to you. I think it'll be the I call you with a brief. And then two hours later, you're showing me a bunch of different options and saying, which ones of these do you want me to go through?

18:55And four hours from now or six hours from now, you might be done with that particular job. And so the nature of the work changes. And if you say, I'm a graphic designer, but I don't know how to use AI tools, then you will be under tooled. It'll be a little bit like saying, I'm a graphic designer. I don't know how to use Figma or Adobe. I have a friend who is a graphic designer who last week was made redundant without any kind of training in AI or anything. They've just kind of said AI has replaced you and he's lost his job. And it's really interesting because he now genuinely doesn't know what to do.

19:25And so I see you totally see a point of everyone's got to be, if you want to be part of this, you've got to understand AI and you've got to use it in your job as much as you can. But there is also the sense of companies just going, sorry, it has replaced you, goodbye. And by the way, companies will naturally start that way. Companies drive their efficiencies is how do they reduce cost? And so they'll go, oh, I can reduce the cost here. I'm going to do that. That's a natural thing. It's part of the efficiency and productivity that companies operate on. There's an individual prescription and there's government prescription.

19:56Individual prescription, you know, he or she should start learning AI tools. They're freely available, you know, here in the UK. And then say, hey, I now have an edge because a lot of people haven't started using those tools yet for it. I now know those tools. And by the way, AI is really good at teaching. One of the things I use AI for all the time is teaching me things so I can be learning things. Like what? What's it taught you recently? Well, recently, I would say that because I was talking to a friend of mine about water pollutants and what to do about it, we used deep research to get a whole report on it.

20:32This stunned my friend about how good the research was carefully. We kind of looked at it. It's like, what are the places that AI is going to need to improve to even do this better? But basically took the report away to be the baseline of the report that they were going to produce for the French government. Right. That was like an instance of a personal thing. But by the way, yours was saving lemon trees with coffee grinds. I mean, like now you know this and you've learned it. So it is learning across the entire thing. And by the way, if you say, hey, I need to be acting more like a tutor, you can put the prompt in.

21:02Don't just give me the answer. Lead me to the answer on this. And if you put that prompt in with what you're doing, then it will start doing that. But should the onus be on the person to do that rather than there be help? Because there's lots of people who wouldn't even know where to start with AI. There's a communication issue around it, the education element to it. I think the answer is all. I think part of the reason I start with, hey, you as an individual can be self-empowered. I mean, literally, if you go to ChatGPT now and say, I want to learn this, help me learn this, it will go great. It won't just try to give you the answer.

21:32It'll try to help you learn it because it's helpful that way. Now, does that mean that that's all we should do? No, I actually think it would be good for companies to be saying, hey, look, how do we engage with our employees to help them learn this new thing? I think that there should be programs there. I'm sure there are startups trying to offer educational reskilling to companies, you know, kind of saying, hey, here's something you could do for new skill training around with AI. And just to get to the government point, this is the kind of thing that government should be thinking about. It's like, OK, how do we make sure AI isn't just the transformation challenge and the huge productivity increase, but it can also help us with the transition?

22:12So like, how do we make sure that there are AIs that are, your friend has made redundant that says, hey, we have this AI assistant to help you figure out what jobs you can possibly do, how to learn the new skills of the job and actually how to do those jobs. And I actually am strongly of the belief that, for example, something that are skilled like graphic design, what happens with AI is the fact is you say, hey, I was really good with my physical dexterity, my hands, you know, that becomes less relevant. But the visual imagination, the thing about how to do that, that's still, if anything, that's too X relevant.

22:47And so it's like, oh, I still have that really unique skill. I'm going to be applying that to graphic design and people are going to be using me for that. Yeah, because people don't understand quite often that their skills are transferable. They're just kind of siloed into one particular industry, aren't they? As an optimist about this, you believe that rapid diffusion, rapid adoption will give a particular economy an advantage. but if you just look at the history of digital over the last 20 odd years, one of the characteristics of these digital networks is it is winner takes all and we have seen the most extraordinary fortunes accruing to a small number of people.

23:37We talked earlier on about Elon Musk, richest person in the world, definitely a beneficiary of the way the earlier stage of this industrial revolution has generated just super normal profits for small numbers of people. Are we going to see in this industrial revolution a worsening of that kind of inequality, much more power accumulating not just to billionaires, but multi, multi, multi billionaires. I do think that inequality grows as the equivalent of network density grows. It actually grows also with, you know, when you have global trade routes and you have kind of globalization because you can now have a product that can ship everywhere in the world.

24:24And of course, that's even more with digital because with digital, if I have the thing that is five or 10 % better that I'm offering through the internet through mobile, then much higher percentage of people will adopt it because it's kind of like if people need an illustration, it's one of the reasons why Amazon and Walmart, when they spread into places, it creates problems for small businesses because they can use their scale of offering these products. And then the small business doesn't have the ability to compete on that 10 % margin of price or price to quality and that kind of thing. And so, I mean, they have a location, which they try to hold on to as a function, but that's a general challenge.

25:05And the digital is the most extreme. Now, I don't tend to deeply focus on reducing inequality because, you know, like, for example, people say, well, the CEO should not make more than 20 times what the lowest person is paid. I don't have any magic numbers like that. What I try to do is I try to make sure that kind of the bottom quartile of people and up are all elevated and given economic opportunity made to have better lives and so forth. Then the gap between the bottom quartile and the top quartile or the bottom quartile and the top 10%, I'm not as focused on that that must be shrunk unless there's particular things like issues around how society may work.

25:44What I'm focused on is how do you get everything elevated. Obviously, we all want to see the vast majority of people better off. And I think if AI can do that, that's great. But if you look at the extraordinary power that accrues to a Musk with his enormous fortune, if you look at the way that the Washington Post has become a sort of plaything of Jeff Bezos, you know, basically ordering the opinion bit of the paper to adopt a completely different set of values from the Washington Post's traditional values. Doesn't that degree of personal power that stems from super wealth, doesn't that worry you?

26:28Well, look, part of, I think, the thing we do in human society, human society, there is power, and there are people with a lot of power. And what I try to orchestrate is that you have multiple powers that are held in balance with each other is a company parallel. If you have one company that's a monopoly, it's a serious problem. If you have 10 companies that are competing with each other, they're held in balance. And that's the kind of thing that I try to look to do. Now, is there examples where power is misused, including power by wealthy business individuals? Yes. But it's also the case where, you know, like, for example, I think the power of celebrity is misused.