Transcription
Thank you. And Rishi, welcome to the Bloomberg New Economy Forum. It's exciting to speak to you. Obviously, you have, you were former Prime Minister of the United Kingdom, but you also have a background in tech. You went to Stanford in California. You've had a deep interest in technology, and you're now an advisor to two Anthropic and Microsoft. So a lot to learn from you about this kind of interplay.
But let's start with a big question, the 10,000-foot view. Everybody's talking about the global AI race. That's kind of how it's being framed, and winning it, and it being a battle between the US, China, and Europe, sort of overseeing, perhaps from a regulatory, regulatory perspective. What does that even mean to win the race? Can you define that?
Yes. Well, morning, pardon me. Morning, everyone. Great to be here and delighted to have this conversation with you. Or is this something that you've thought about deeply as well? Now, you're taking a step back before we talk about the AI race. I think, you know, we're living at a time which is both, paradoxically, the most dangerous of our lifetimes, but also the most transformational. I say dangerous because the old unipolar world order that we were used to is gone. We're not quite sure what's replacing it. You know, great power competition is back. Technology is obviously a part of that. But also, this is transformational because I, you know, I genuinely think will be a general-purpose technology, like electricity. Yeah, like, like steam. And yeah, these things don't come along that often. And when they do, they have the potential to transform societies, economies. And that's really exciting. Yeah. So that, that's, that's where we kind of situate ourselves at this, at this moment.
Now, when it comes to AI rights, there is all this focus on the race to agree, who's going to get to the singularity? Is it the US? Is it. Is it China? I actually think that's actually the wrong race to focus on. Okay. And I actually think the race we should probably pay a little bit more attention to is, is the the race for who can adopt technology and AI in particular, the fastest. If you look at the history of general-purpose technologies, and actually, there's a professor called Geoffrey Ding who's written about this, and I found his book very powerful. You know, what history tells us is you don't have to be the country that invents the technology to be the one that disproportionately benefits from it. And you think all the way back to the printing press, invented in Germany. Actually, it was the UK and the Netherlands which really capitalized on that innovation. And similarly with the Second Industrial Revolution, even though Britain did a lot of the innovating, America did a lot of the commercializing. And I think that that's the interesting thing to think about, so that who can sprint to the finish line of getting this AI spread across their economy, across their society in a way that benefits their citizens. You know, a little bit more focus on that race, I think would be beneficial.
So diffusing the technology to to people is part of winning it. And it sort of makes sense when you think about how much money is being spent on building infrastructure, that if Europe or the UK capitalizes on, that they don't have to spend that kind of money and just build on top. I want to ask you about your role in in AI. When you were Prime Minister, you established the AI Safety Summit, the very first one in 2023 at Bletchley Park. Since then, there's been a kind of noticeable shift in tone, I think, among governments, or from let's let's oversee AI and make sure it's used properly, to let's adopt it as quickly as possible. Do you think that's that's a correct shift? And do you think there's a possibility that actually there could be a price to pay if governments focus so much on just implementing this new and untested technology so quickly?
Well, I don't think the two things are incompatible, and I think it's important to focus on on both. Now, you know, we touched on the the diffusion point. I think it is important that governments are building what I call the diffusion infrastructure. You know, what does it take to get the technology deployed across your country? It's about training people, training small businesses, bringing public opinion along with you, making sure that you in the public sector are actually doing your bit to drive innovation and transform people's lives that way. So I think that is a really important focus for government.
Now, look, along alongside that, you know, government's first responsibility is to ensure the security and safety of its population, of its citizens, and that's the first order of business if you have the privilege of leading a country. And obviously, when a new tool like this comes along, yes, enormous potential for good, but also potential to be misused. And, you know, my view was it was important that governments develop a technical capability to evaluate those risks and and talk about it with with peers. And that was the genesis of the security initiative and summit that I held. And I'm really pleased that that has now become a fixture on the international calendar. So South Korea did the next summit, France did the summit earlier this year, although the name changed. Yeah. Which I think is I think it's completely fine. It changed to the action summit that I say, and I think India's summit this next year will be the impact summit. And I think that's right. I think the more important thing is that, you know, leaders are coming together every year to focus on. I think I think having a dedicated event where world leaders have the ability to to talk about it, paying attention is absolutely the main prize here. And now that has been firmly established, as I said.
But but I think that's the important point here, is the right thing to do is is in my view, not to regulate. And it's developing the technical capability to evaluate risk. And we set up the AI what is now called the AI Security Institute, which has been rightly mimicked around around the world. And I see I see Gina sitting there and her and I worked on this together when we were both in government. And I think governments developing that expertise to work with these labs to evaluate the risks. And that does have to involve your national security institutions as well, is the right thing to do. Yeah.
And it's fortunate that the tech companies who are building AI models like OpenAI and Anthropic and Google are working with the AI Security Institute to voluntarily give them access to their models. But what happens if those companies decide they don't want to give that access? And for example, Mark Zuckerberg, who is making open-source models with Llama, recently has been saying actually, we'll probably go more closed-source as our models become more capable, as they get closer to that threshold of building machines that are smarter than humans. Doesn't that mean that maybe the AI Security Institute won't get that access it needs to test those models? What do we do then, if they become more opaque?
Yeah. It was so. So far, we haven't reached that point, which I think is positive. And what I've seen for the most part is, you know, companies committed to, you know, what they would call responsible scaling practices and working really well with our institute in the UK, the Institute in the US, and others to do what's called pre-deployment testing. So before the models are released, go through that red teaming and those exercises. And I think actually what those companies have found is they benefit from that interaction because ultimately, they don't have the same national security expertise that exists within governments. So having those people be involved in the process of the collaboration. Exactly. That collaboration is valuable to the companies. I look, I don't think anyone wants to put something out there into the world which they think could genuinely cause significant harm, whether it's cyber, biological, or nuclear or chemical, they don't tend to put it exactly. So if we have the ability to look, let us work through those things with you in a collaborative fashion. And this is not regulation, this is just a technical evaluation, but just to make sure that there are no unintended consequences with algorithms that are perhaps designed for engagement the same way social media put out technology meant to connect us. But it did end up having these harmful consequences. And that's and that's exactly the point.
And look, these companies all themselves conduct this testing, as they should. But I said, I don't think you can as a government just say this has got nothing to do with us. Of course, of course, you need to be able to make those evaluations yourself and add something to what the companies themselves are doing. And so far, that has been, I think, a mutually beneficial and positive collaboration. Certainly in the UK, we found those conversations with the companies have been positive, productive, and they are committed to continuing to engage. And long may that continue?
I was going to say, long may that continue, because that is a great place for things to be right now, and let's hope that that continues to be the case. Let's just talk about something a bit broader, which perhaps a lot of members of our audience can relate to, and that's the job market, and particularly preparing our children for that job market. I have a 15-year-old, you have a 13 and 14-year-old. We were talking about being parents of teenagers backstage and what that entails. With so many university graduates now struggling to get jobs because companies are using AI for intern-style roles, how should how do you prepare your own kids for that? What kind of skills do you think are most important for them?
Yeah, I think I have to kind of teenage young teenage girls. So they'd much rather talk about skincare products than they would. Talk about, you know, preparing for the job market. But I mean, the way the way I've thought about it a little bit, and this is actually informed as well by looking at the work that the Stanford economists are doing, where I spend some time, and also looking at the data on LinkedIn, which I think is a fascinating insight into what's happening in the job market. So you're just looking at LinkedIn yourself, just looking for and seeing what's happening, what you know, what what you take away from that.
What I take away from that and what I tell my girls is that first and foremost, of course, you just need to be familiar with how to use these tools. Yeah, I think that I think everyone accepts that, you know, you see jobs requiring some degree of AI literacy doubling year on year in the UK, in the US. I think that will continue. And that's not software experts or computer scientists. That's just whatever job you're doing, having the the AI literacy skills so that you don't lose your, you know, lose your job to someone who is proficient with them, I think is critical. So I think that's the kind of overriding piece of advice.
I think the few specific things I think about is that we're never going to lose the importance of being able to think, to reason, to question critically. So I think those skills will be incredibly important for our young people to develop. And similarly, those human skills, actually, if you look at the LinkedIn top-ranked skills, you know, only one or two of them relate to AI. The rest of them are all things that talk to human-to-human interaction, interpersonal dynamics, empathy. And so I'd make sure that my girls were making sure that they spent time developing those skills as that as they grow up.
And then I think the last thing which is going to be a real change actually, is everyone, whether you're a new graduate or a senior leader, is going to have to get used to managing a team of agents. And I think that will be, yeah, the age of agents will come relatively quickly. And it's great. You know, we're used to learning how to manage teams of people, but I think actually very young graduates will soon have to learn how to manage a team of agents and understand what that means, how to divide up tasks, how to make sure that the work being done is is right. And that will be a new skill. And I think, you know, the nature of work is going to change in that regard. And, you know, my kids, your kids are going to grow up into that world, and I'm excited for them. But it will require, I think, a mindset to just keep learning and keep curious, being able to make, to adapt, to adapt. Yeah.
Well, Rishi, honestly, I have so many more questions I could ask you, but we are unfortunately out of time, so we are going to be bringing on our panelists shortly. But for now, thank you very much, Rishi, for that so much conversation to take.