Dambisa Moyo is a member of the United Kingdom’s House of Lords, a long-time board member at large international companies, and the author of five best-selling books on economics, geopolitics and business management.
Baroness Moyo, a native of Zambia, is also a global economist who lives between the UK and US, and travels extensively. Her preferred method of research is in person.
She has been a vocal advocate for the benefits of AI for society while warning about its potential to widen income inequality. When the Review spoke with her during a recent California visit, she told us her research in recent weeks has shifted her outlook on AI.
In a 2023 essay for Project Syndicate, you wrote that “there is no reason to think that AI will not” benefit society in the way that technological achievement traditionally has. You foresaw possible enhancements in trade, global connectivity, education, public health and more. Two years later, are those benefits as likely as ever?
Maybe not. I’ve had numerous conversations with people recently that have surprised me. Essentially, there is a concern about compute power; they feel the speed and gains of LLMs may be stalling. I think if you’re not in Silicon Valley, you’re assuming there’s more money getting pumped in, and that money is going to drive more hyper-scaling. But I’m hearing a lot of people talk about hard constraints around chips, data. That there is no more exponential growth in knowledge and speed to be expected than what we’ve already seen.
There are clearly AI gains to be had for individual consumers. I can search faster and more comprehensively; I can use ChatGPT to get information. But there are still too many errors—hallucinations, as they’re called—for AI to be fully rolled out at the enterprise level. It’s one thing if AI makes a mistake while you’re searching for information—it’s another if that mistake affects an airline’s entire ticketing system.
From here on out the gains are going to be more incremental?
Correct. Some say the next burst of a big gain could be 50 to 100 years from now. I don’t know. Investors I’ve spoken with have said that progressive large language models just cannot defy the laws of physics and produce a faster AI.
I spend a lot of time talking with the top venture capitalists. From an investment standpoint, nobody is generating attractive returns. Somewhere between $60 billion and $130 billion is going into AI in venture capital every year, with massive expectations about what the returns on these investments should look like. Everyone invests believing they are going to see returns like Nvidia’s. But the smartest people I speak with in venture capital say there’s a very high likelihood that most of these investments are going to zero—that is a lot of misallocated capital.
“Somewhere between $60 billion and $130 billion is going into AI in venture capital every year, with massive expectations about what the returns on these investments should look like.”
If the benefits of AI are slower coming than expected, would that be true as well for the real-world risks?
I don’t think the risks are gone. Everything I’ve been hearing has been focused on the US. I certainly believe that there are equally smart people in Russia or China sitting around a conference room debating these issues, but I have a less clear sense of what the risks are like there.
In terms of what I think I do know, in terms of what we as a society think we know, I always think about that adage attributed to Mark Twain: “It’s not what you don’t know that gets you into trouble, it’s what you know for sure that’s just not so.”
In your July AI column, you wrote that 2023 R&D investment among the “Magnificent Seven” Silicon Valley giants equaled the entire R&D budget of the European Union. Is that worrisome?
What’s worrisome is the concentration of power. If it becomes the case that a handful of companies have more data, more information, more influence in an AI age than a region or a country, how are we going to react to that? I think we’re going to start asking some really hard questions about centralization of power, ownership of models and data.
Bitcoin, in a way, embodies that. In December you saw it trade at over $100,000. People can dismiss it, but at some point it becomes too big to ignore. All the talk about how governments won’t allow bitcoin to exist because governments derive their power from fiat currency—that’s going to be challenged. The balance of power is shifting dramatically. And I think governments will struggle to fight back.
It’s not just power. It’s talent. When asked recently, a president in Europe said their biggest concern is how quickly they’re bleeding talent. That talent is going to places like the US, and Silicon Valley more specifically.
My main point is that there are new avenues of power and influence emerging with AI. These new avenues are further concentrating power, influence and wealth. It’s moving so quickly that it’s leaving all the other forms of traditional power, the way we’ve traditionally distributed information, behind.
You’ve written about the danger of workers getting left behind. Who is responsible for making sure that workers benefit from AI—government, industry, or both?
I’m a believer in market capitalism. I see capitalism as the most innovative, biggest driver of progress. And my response is, it’s government’s responsibility. Some libertarians may call that government overreach.
But I don’t think corporations are going to solve this problem. They may wish to solve this problem. They may make efforts to solve these social problems. But I think when all is said and done, it’s government that has vested interest in seeing a more equitable, more progressive society. I don’t think that the nature of corporations makes it possible for them to do that.
How might government act? You’ve written about the possibility of a universal income.
I think everything’s on the table. It’s very eye-opening for me to travel around the world and see how other people are living, and how disconnected their expectations are from their realities. And I think governments need to be much more aggressive about trying to close that gap.
Because that’s what revolutions are made of. I’m not being a catastrophist; I’m simply saying we’re in a period of low economic growth, higher inflation, and AI that could potentially continue to displace routinized work. It seems to me that governments need to offer something to stabilize that situation. I just published a paper on the eight economic headwinds facing the globe in 2025.
Right now everybody’s betting on the US, and it’s the only place where there seems any prospect of growth, enthusiasm or innovation. Everywhere else I’ve been is feeling grim and I don’t understand how a unipolar economic juggernaut survives with the rest of the world in as much chaos as it’s in. To be in the US is comforting, but it can make you delusional about the challenges going on in the rest of the world.
You wouldn’t know that from the Trump opponents who talk about moving to some
better country.
I’d love to know where.