Parmy Olson, AI expert: Who controls the future?

0:00This BBC podcast is supported by ads outside the UK.

0:30Singapore, where business events can create lasting impact. At the BBC, we go further so you see clearer. With a subscription to BBC.com and the BBC app, you get unlimited articles and videos, ad-free podcasts, the BBC News Channel streaming live 24-7, plus hundreds of acclaimed documentaries. From less than a dollar a week for your first year, read, watch and listen to trusted, independent journalism and storytelling. It all starts with a subscription to BBC.com and the BBC app. Find out more at bbc.com slash unlimited. Hello, I'm Misha Glennie and this is the interview from the BBC World Service.

1:13The best conversations coming out of the BBC with people who are shaping our world from all over the world. If you're not a little bit afraid, then you're not paying attention. We have never seen a people so united. Do not make that boat crossing. Do not make that journey. Being born in America, feeling American, having people treat me like I'm not. We're more popular than populism. For this interview, I speak to AI expert and writer Parmy Olson about the battle for supremacy at the heart of the artificial intelligence revolution and how this fast-developing technology is going to change our world.

1:56She challenges the claim by tech leaders that AI is a democratising force, putting powerful capabilities in the hands of people. Instead, she believes it can reinforce inequality, with giant technology firms now valued in the trillions of dollars. Such is their financial status, she says, it's now increasingly difficult for governments to regulate them effectively. These companies don't really, they don't care what governments do. Like their priority is their shareholders and is their own existence and their own, you know, the next quarter of growth. And I know there is the possibility that governments can regulate them, but they just don't.

2:39They don't regulate them properly. The most they get are, you know, multi-billion dollar fines, which sounds like a lot. But actually, in the grand scheme of things, it's pocket change. You know, it's like a parking ticket for these companies.

2:55Welcome to the interview from the BBC World Service with Parmy Olson. How important is AI going to be for the future of the world? I think it's going to be very important, but I think it's going to be hard at first to see the impact. And I think one way to think about it is the Internet. when people in the 50s and 60s, 1950s and 1960s, tried to imagine what the future would look like. They imagined robot butlers and flying cars. And of course, we don't have that, but we have something that was just as monumental, if not more so, which was the web. And it created this entire economy. It changed the way people speak to one another.

3:43It connected people globally in ways we couldn't imagine. And I think AI will be similar in the sense that if we look 20, 30 years from now, we'll still be walking around in the rain in our coats, as I was just now, looking at our phones on the train. But we'll be thinking differently and our culture will be changed and our human relationships will be different and our jobs will probably be very different as well. So I think there will be a big disruptive impact on jobs. I think many jobs will get lost. We're seeing this already in the creative sector. I just saw some stats recently that 58 % of photographers and two-thirds of illustrators say they've lost work because of AI.

4:26And 86 % of authors say they're earning less money now because of AI. So, you know, these are detrimental effects. But another way is there are positive effects. If you're an entrepreneur, this is a fantastic time to start a business because you can get your invoices automated. You can get legal advice. You can get all sorts of administrative help that would have cost a lot of money in the past. And so these are the kinds of impacts that I think there'll be lots of second and third order effects. What happens when you suddenly have more people building businesses? What happens when inside a company that is building software, instead of having three people on a team, a designer, a marketer and a programmer, you just have one person?

5:15Both the proselytizers of the Internet and AI promise that this will take us into a land of plenty. And yet what we're seeing in the age of the Internet and now in the age of AI is actually growing inequality in wealth and social status around the world. is there going to be any point where people say enough is enough we need a better distribution of resources than we're getting now yes um i you know again if there is some kind of popular uprising let's say who do you protest against do you protest against google or open ai I, the tech companies and the tech leaders would argue that they are, quote unquote, democratizing this incredibly powerful technology, putting it in the hands of people in order to empower them to, as I said before, build a business, be more productive.

6:22So I think, yes, I completely agree with you that we are moving in a direction where this kind of technology can actually make inequality so much worse. Because when we look at who truly is benefiting financially from generative AI, it's the world's largest tech companies. I mean, you can just look at their market capitalizations, for example, over the last three years. Those numbers have gone up by something like six to eight trillion dollars for the six biggest tech companies. That's not even how much they're worth. That's just how much they've gained in market value. You know, companies like OpenAI and Anthropic are able to raise hundreds of billions of dollars, some of the biggest fundraising rounds in history.

7:10So yes, a lot of people are becoming enriched and powerful. And there's a real question about how those financial benefits are trickling down, as it were, to the rest of the public. It's not clear yet. Let me project 20 years ahead. The US and China, they're competing with each other for superiority in various technological sectors. But let's say one has a significant advantage in AI. Is that going to give them a decisive advantage in the battle for geopolitical control around the world? I sometimes wonder if the element of geopolitics is overplayed a little bit with technology. Yes, I suppose you would have the upper hand.

8:00But I also just think these companies don't really, in a way, they don't care what governments do. Like, they are, their priority is their shareholders and is their own existence and their own, you know, the next quarter of growth. And I know there is the possibility that governments can regulate them, but they just don't. They don't regulate them properly. The most they get are, you know, multi-billion dollar fines, which sounds like a lot, but actually in the grand scheme of things, it's pocket change. You know, it's like a parking ticket for these companies. And so I think, yes, there are bragging rights, I suppose.

8:39And it's interesting how different countries frame what they see as winning the AI race. Like, what does that even mean? For Britain, they see winning that race as a race of adoption. So if the British public and British businesses can adopt AI more quickly than the Americans or the Chinese, which is very unlikely because China's way ahead on adoption as well. But let's just say that was the goal. Then maybe maybe Britain could win the AI race. So it's interesting because in your book, you focus less on that US-China competition and more of the battle between the Silicon Valley giants. First of all, the paradox of these tech visionaries who wanted to build artificial general intelligence with this very deeply utopian vision of making the world a better place.

9:39It's a little bit of a cliche to talk about tech bros wanting to save the world. But these guys really did have a messiah complex. I'm talking about Sam Altman, the founder of OpenAI, and Demis Hassabis, the co-founder of DeepMind. And Demis was actually the first one who wanted to start a company expressly for the purpose of building artificial general intelligence. No other company like that existed before 2010 when he and his co-founders did that. And so when ChatGPT came out in November 30th, 2022, I remember thinking to myself, there's so much more to this story than just this AI, incredible little AI bot that is probably going to change things.

10:29There has been this effort to build AI for the betterment of humanity, and that effort has been co-opted by enormous corporate interests. And both these visionaries had been worried about that. Both of them had tried to create governance structures that would protect this technology, this AGI they were trying to build, so that it wouldn't fall prey to vested interests, corporate interests or a single person's financial interests, but would be guided responsibly. And actually, both of them failed to set up those structures and they had both kind of sold out to big tech in crude shorthand.

11:16You're listening to the interview from the BBC World Service.

11:25It's 2009 and we're in the German mountains. A man straps himself into a car on the world's most dangerous racetrack. He whispers to himself, It's time to put my balls on the dashboard. As he starts the engine. In 15 minutes, he's in an ambulance, unconscious. In 15 years, he's a billionaire. This is Total Wolf, Formula One's most powerful team boss and the breakout star of Drive to Survive. This week on Good Bad Billionaire, how Toto Wolff made his billions. Listen wherever you get your BBC podcasts. So for many years now, I've been giving talks on cybercrime and cyber security. And when I give those talks, I often find myself staring out at an audience a sea of men.

12:10Recently, however, I've noticed that more and more women are writing about and engaging in technology. This is very important because technology impacts on women in very different ways than it does men. And when I was talking to Parmi Olsen, I was particularly struck about how she analysed not just the technology, but the sociology of technology, and in particular, the people, men mainly, who run these industries. It's quite eye-opening in my opinion. So let's return to that conversation with Parmi Olsen. How do you assess China's AI capability? China has moved incredibly quickly on AI. And one of the reasons is this real embracing of open source in the country or what you might call open weight models.

13:03So these are models that are free to use and free to copy and free to develop on. And this means that tech companies in China, when they're building a new AI model, they don't have to start from scratch because their competitor's AI is not proprietary. They can just take that model and build on top of it and make it better. So it's been almost a kind of collaborative effort of companies in China that has held sort of a tide that has raised all boats, I suppose you could say, for the ecosystem there. And now we're at a point where there's a little bit of disagreement about how good AI models are.

13:42Funnily enough, people in China will say they're still behind Silicon Valley. You talk to technologists in Silicon Valley and they'll say they've caught us or they're almost ahead. It's a bit of both. If you look at big corporations who are using AI models today, they're primarily using the likes of OpenAI's model, Google, Anthropic. They're using the Silicon Valley technology. But look at startups, smaller businesses in the U.S., even in Europe. They're using open source models from China because they're cheaper and they're just as capable. If you look at some of the rankings of the best large language models available today, models like Alibaba's Kuen from China, DeepSeek, Kimmy is another one.

14:32They're right up there with ChaChipiti and Gemini. Provided you don't ask them whether Xi Jinping is a good president or not. Yes, there's a little bit of that, I'm sure. So how important do you think in Silicon Valley was the launch of DeepSeq around the time of Trump's second inauguration? Did it spook them? The launch of DeepSeek caused a freakout in Silicon Valley because here was a model that was as good as ChatGPT, but built for seemingly a fraction of the cost. And it suddenly raised these questions. Well, OpenAI and Google and Facebook and Anthropic, they're spending billions of dollars on all these data centers.

15:16Is that just a waste of money? And so in January of last year, gosh, we're kind of in the anniversary of that moment right now, shares in companies like NVIDIA, I think NVIDIA dropped like 10 % at one point. And of course, the stocks recovered because there was an understanding that actually businesses were still buying the Silicon Valley products. But I think there was a lingering sense still. But, you know, China is catching up. It's not just DeepSeek. There are more companies just like it that are coming. And it strikes me, you mentioned NVIDIA. NVIDIA is one of the few companies, along perhaps with TSMC, the microchip manufacturer based in Taiwan, that actually make any money out of this.

16:02The rest of the companies, you know, the chat GPDs and DeepMind and so on, don't appear to be making any money. Is that going to change at any point or is it just going to be NVIDIA laughing all the way to the bank? Well, I don't think NVIDIA is in necessarily the strongest position either. It kind of, you could say, has all its eggs in one basket, right? It has a few big customers, and if anything happens to them, that's a problem for NVIDIA. I think that's why NVIDIA is trying to move into robotics so that they can diversify a little bit. I would also push back a little bit on some of these other companies not making money.

16:39Anthropic actually has a thriving enterprise business. It is selling AI models to businesses and it's making billions of dollars. I think – sorry, don't quote me on the run rate, but it's – I think it was six billion. Yeah, better not quote me on that. But companies like Anthropic and even OpenAI, massive financial constraints. Sure, they're on the hook for something like – this is crazy –$1.4 trillion in computing commitments, which is something like 100 times more than the revenue they made in 2025. So in order to meet those obligations, they have to make tens of billions of dollars over the next few years, hundreds of billions, really.

17:24So how are they going to do that? Well, they're going to sell their AI models to businesses. That's one. But Sam Altman also just turned on the spigot for advertising on ChatGPT. Now, advertising is one of the most successful business models of all time, and it's what kind of keeps the internet going. It's why Meta has become a multi-trillion dollar company. 98 % of their revenue comes from ads. Google, it's 80%. You know, ads can make a huge amount of money. So I see there certainly being a possibility that open AI could get close to meeting those obligations. It could make a lot more money. And Google as well.

18:07They have incredible distribution, just like Microsoft. You mentioned Great Britain and its approach to AI. How is Europe doing? Is Europe lagging behind everything or are there some encouraging developments there? Europe definitely has some encouraging developments. We have Mistral, which is an AI model maker, open source AI model maker in Paris, and Hugging Face, which is another open source AI model company. And Jan Lecun, who was known as a godfather of AI because of his contributions to deep learning, left as Meta's AI chief, said goodbye to Mark Zuckerberg. And he's starting his own company in Paris as well to build world models.

18:57So Paris is really becoming a hub, I think, right now for artificial intelligence. London as well. There's a lot of AI startups here. We've got Wave, which is a self-driving car company. We've obviously got DeepMind. Of course, it's part of Google, but you have a lot of people leaving that company and starting their own businesses, and they're able to raise a lot of money. On that point, this is where Europe is really struggling. It's raising money. If you're a startup and you want to raise 20 million pounds, 100 million maybe, you can probably do it. You want to raise more than that, you're going to really struggle.

19:35So this is a well-known issue in the UK and across Europe. We don't have a startup problem. We have a scale-up problem. And pension funds in this continent, in Europe, are just not spending on new technology and on startups. So if you look at the pension fund assets across Europe, it's something like 10 trillion euros. The proportion of that money that goes to venture capital, which is like the funds that fund startups, it's 0.018%. It's tiny. It's like almost nothing. If you look at the same proportion in the U.S., it's 2 to 3%. So that sounds small, but it's actually orders of magnitude more.

20:20We're talking oceans of capital that are available to startups in the US because of that. So often tech companies like Meta, like Google, will complain about regulations holding back tech and startups here. But I don't think that's true. I think that's a red herring. The real problem is the lack of funding by pension funds, by the financial districts in London, in Paris, in Amsterdam, into startups. You've written about quantum technology. how far away are we from a from a feasible quantum computer well we have quantum computers now i think the question is can we actually build them at scale and is it possible to do them cheaply so that you know lots of banks can have them i know hsbc recently said that they used a quantum computer in quite a financially beneficial way through some bond trading exercise.

21:19But I think it really depends on who you talk to. If you talk to people who work in quantum computing, they will tell you that it's imminent. And that's the same kind of rhetoric you hear from people who work in AI. AGI is imminent. But the real answer is nobody knows. But when it does happen, and by the way, this is another area of technology where China is doing really well and is catching up, I think there's going to be some real questions about the security of our data, which is not the first thing you think of when you think of quantum computing. You think, oh, this is going to be a huge impact on the markets and we'll be able to kind of do all these tasks and run AI models much more quickly.

22:02But actually another kind of second-order effect of that will be any kind of encryption that you have protecting your financial data, your personal data, that can very easily be broken now by a quantum computer. So that has real geopolitical implications. Oh, yes. That is something. See, it's funny you say that, because when we were talking about artificial intelligence and the geopolitical implications, I see the kind of global trade of ideas and open source models and scientific research on AI as just being a little bit more fluid. whereas with quantum computing there are huge security implications.

22:42And so for me the geopolitical fallout from that is much more obvious and meaningful. What about energy environmental considerations? We've heard about data centers soaking up huge amounts of energy, water. Do we have even the capacity to generate this amount of demand? It's interesting. Whenever I talk to teenagers, I've been to a few schools and talked about AI to secondary school students. And whenever they ask me questions, the number one thing they always ask is about the impact on the environment. They're really worried about that. But I'm actually less worried about it than maybe I should be more worried.

23:29But having spoken to people in Silicon Valley who work for Google Cloud, who work for Amazon AWS, for these people, it is in their financial interest for these data centers to not be so power hungry because it's really expensive for them. So they are working very, very hard to find more efficient ways to design algorithms, more efficient ways to use all that computing power, the power from those chips. And I think when the incentives are set up like that, my hope is that they will find a way to do this more cheaply and more efficiently.

24:12Thank you for listening to the interview. If you enjoyed this conversation, you can find many more episodes of The Interview wherever you get your BBC podcasts, including ones with Google boss Sundar Pichai and Mustafa Suleiman, CEO of Microsoft AI. Until the next time, bye for now.

24:44At the BBC, we go further so you see clearer. With a subscription to bbc.com and the BBC app, you get unlimited articles and videos, ad-free podcasts, the BBC News Channel streaming live 24-7 plus hundreds of acclaimed documentaries. From less than a dollar a week for your first year, read, watch and listen to trusted, independent journalism and storytelling. It all starts with a subscription to bbc.com and the BBC app. Find out more at bbc.com slash unlimited. You