Artificial intelligence promises to transform the economy, boost productivity, and expand human capabilities. But fears of mass unemployment, catastrophic risks, and the rapid expansion of data centers are fueling calls for restrictions that could slow AI development and delay those benefits.
In this episode of the Human Progress Podcast, Kristian Stout, Director of Innovation Policy at the International Center for Law and Economics, joins Chelsea Follett to explain the costs of slowing AI innovation and the obstacles already standing in its way. They discuss why AI could create more jobs than it eliminates, how excessive regulation could leave the United States dependent on Chinese AI technology, and how conflicting state laws, local zoning restrictions, and infrastructure bottlenecks threaten to slow the industry’s growth.
Below is an edited and abridged transcript featuring some highlights from the interview.
Joining me on the Human Progress podcast today is Kristian Stout, Director of Innovation Policy at the International Center for Law and Economics. His work addresses competition, telecommunications, and artificial intelligence policy—very big topics these days.
Let’s start with a big-picture overview. Can you sketch out the history of AI and data centers for us before we dive into the current debates?
AI feels like something very new. When the computer starts talking to you, that’s unusual. But it’s really part of a longstanding trend in computer science. As far back as the 1940s, there were researchers trying to understand how you could model, for instance, human neurons inside of computers.
So, the whole project of computing is trying to introduce more intelligence to computing, and what we have today is the result of compounding developments over many decades. In the 1970s, you started getting some more sophisticated algorithms being deployed. The ones that consumers are most familiar with are things like autocomplete and search engines. Those are all forms of artificial intelligence.
Then, around 12 years ago, we started really focusing on machine learning. For a long time, the news was about “big data.” All of the investment money was going to big data and different machine learning techniques. That was artificial intelligence. Nobody was writing news articles about it as artificial intelligence. There were some anxieties surrounding it about, for instance, what that augmentation meant for labor and the future of work. But we didn’t frame it in terms of Skynet taking over. Today, because you can interact with the computers in a way that feels very human-like, it’s become popular to think about AI that way.
It’s a similar story with the data centers. Data centers have been with us since the 1950s, albeit at a very small scale. IBM’s mainframe model, where researchers would share information over long distances with phone lines, was an early form of data center. The data centers that we think of now as “the cloud” really started being built around 2005, when the scale of the internet really was taking off, and we needed a lot more computing support. So what we’re dealing with, when it comes to both AI and data centers right now, is a scale difference, not a kind difference.
That scale difference has brought with it some interesting political debates. The public’s view of data centers has seemingly soured even as people are using AI more and more. And we’ve been seeing more and more political moves to ban or restrict the construction of data centers.
What is your take on this development?
I think all of your viewers are going to be familiar with the idea that people’s true feelings do not always match what they say in public opinion polls. You will get a different result if you ask something like, “What do you think about data centers?” versus “How do you feel about an industrial project in your area that might bring jobs and might have these costs and benefits?”
It’s the same thing with AI. The CEO of Anthropic, Dario Amodei, goes out and says something about how AI could kill us in 10 years. So when you poll people, “What do you think about AI?” well, “I would like it not to kill me in 10 years.” But if you then had a poll asking something like, “How would you feel about a computer technology that enables you to do all of your week’s work in two days?” you might get a more positive result.
Part of it is a categorical thinking problem. As human beings, we look at the environment and create groups of things that allow us to make very fast decisions. That decision-making process is often not complete enough for us to actually answer complicated modern questions. I think most people, if you actually drill down into the questions with them, have a more nuanced opinion about these technologies.
I think that there are understandable concerns about what AI will mean for the future of work. Personally, from the research I look at and from interacting with these technologies, I don’t think AI is going to cause mass layoffs. I think employment will go up. Workers will be expected to be even more productive, and we will invent new forms of consumption that we need to satisfy.
However, it is very hard to predict what that future economy will look like. AI is what researchers call a general-purpose technology. It’s like electricity. It’s a platform that people are going to use to develop new ways of doing things. So we won’t know exactly what AI will be used for until the market discovers it. That creates some understandable anxiety. But one of the nice things about this particular technological revolution is that it helps us have cognitive adjuncts. AI itself can help us figure out how to onboard more people more quickly and more seamlessly into this new intelligence revolution in a way that previous technological revolutions couldn’t.
You said that you believe with AI, employment will go up. Can you elaborate on that?
If you take human nature seriously, we have an unlimited amount of desire. For hundreds of thousands of years, humans have been looking at their environment and thinking, “I can do a little bit better. I would like to eat more; I would like to have more mating opportunities. I would like to have more of everything.”
There are tons of spiritual traditions around the world that try to help us learn how to tame those desires. So far, they have not been very successful. Instead, when a new opportunity presents itself, our desire fills the volume you provide to it. So if there’s a technology that lets you finish your workweek in a day, I don’t think people are going to stop working on Monday.
Another thing to keep in mind is that we’re status-driven creatures. The reason why we value money or other resources is not just the resources themselves, but also the status that those resources provide us relative to other people. That means there’s never enough. If someone else is perceived as having more status, we’re always going to want to increase our own status.
Modern wealthy people in the United States compete on their investment in their children, right? When I was a kid in the 1980s and 1990s, no one went to anybody’s practices. Now, sports and extracurriculars are a kind of social event where people vie to be the most involved. AI is going to lead to something similar. It’s going to enhance our productivity, but the things that we value are going to expand in turn, and we’re going to fill that time with more value-seeking opportunities. And that might not look like sitting at a desk for 40 hours a week, but it will still exist, and there will still be employment and resource opportunities within those new activities.
Some of the anxiety around AI is inspiring policies that could slow down the industry. There’s a big difference, for example, between Europe’s technology industry and the relatively less regulated technology industry in the US. Could you talk a bit about how the policy environment affects the ability of the AI industry to continue to innovate?
The European area, before it was the European Union, was one of the leaders in pharmaceutical production throughout the 20th century. They’re still developing these drugs and they’re still innovative, but over the last 30 years, because of policy choices they’ve made, America has just blown past them because we have a relatively pro-innovation regulatory environment when it comes to pharmaceuticals.
There’s definitely a similar risk with AI regulation. First off, China is becoming a very innovative place, and I don’t believe they’re going to slow down AI innovation. They’re standing up their own chip producers to get around our export controls, and some of the early returns on their chips look very promising. I myself run a lot of the open-source models that come out of China. They’re fantastic. They do a great job.
In Europe, they’re largely a consumer of our technology products. They don’t need to be. There are brilliant people in Europe. They’re constrained by the regulatory environment. The cost to Europe is that they’re six months behind us, say. The United States gets a product first, and then a team of lawyers has to figure out how to deploy it in Europe. And then Europeans eventually will get some form of it.
That’s a serious cost. I read a lot of science fiction as a kid, and I really loved cyberpunk novels. One of the concepts in those novels was that everyone had personalized AI defensive measures. Your AI is essentially supposed to go out and be your agent, to go out and do the things you need it to do. And part of that is defending you from the bad AIs.
The future could look something like that. You may have an AI agent whose job is to protect you and advance your interests. It might check your identity theft exposure, check your financial risk exposure, go out and make sure that the opportunities in your professional life are being maximized, and so on. If we lose the race to AI because we have bad policy, the cost could be becoming de facto citizens of Chinese AI infrastructure in the future.
I want to get your take on the recent AI Safety Accord. So this is a matter of self-regulation. Some have critiqued it because it’s non-binding. What’s your take on the AI Safety Accord and this idea of industry self-regulation?
There’s no such thing as self-regulation anymore. We do not live in a laissez-faire environment. We live in a highly regulated environment, even in the United States. And regulation is not just someone at the FTC sitting down and designing a code of conduct. The legal system itself is a regulatory apparatus. The United States has a common-law approach, where we have adjudication through different court levels. And we also have pseudo-courts within the regulatory system. That backdrop means that there are liability rules operating in the background of every firm that will potentially destroy your company if you run afoul of them.
But I guarantee you AI companies have counsel inside warning them to be careful. You could see negligence liability emerging based on the way consumers use these products, the way you’re selling them to people. There are all kinds of exposure for liability. It’s only a matter of time before you start to see more settlements and cases emerge.
So by saying we’re encouraging them to do self-regulation, we’re not saying AI companies can get away with whatever they want. We’re saying they need to take the first crack at helping us understand how to understand AI and how it fits in society. And when something goes wrong, a court or regulator is going to look at what their efforts were and whether they were good enough. If they weren’t, AI companies are going to face massive liability.
What do you see as some of the most damaging roadblocks, even in the United States’ relatively pro-innovation environment, that could potentially cost us the AI race?
Great question. There are two broad classes of problems right now. The first one is the state versus federal distinction. The second one would be the data center pushback.
Because the federal government has not been able to enact something like a federal framework for how we want to think about AI regulation, we are seeing a lot of states rushing to fill the gap. If states end up having different requirements for training a model, safety testing a model, and offering it to consumers, that would definitely put the brakes on AI development.
The received wisdom right now is that Chinese models are about six months behind American frontier models. Maybe a little less in some areas. If you were able to introduce enough regulatory friction in the United States that major labs felt like they needed to severely slow down their deployment, you could easily see China match us or even start to outpace us at the frontier level.
The other major roadblock is data center permitting.
One problem is zoning at the local level. A hyperscaler or somebody who wants to build a big data center comes into a community, and the local council is like, “What questions do I ask? What data can you give me?” The public processes that the states have in place can be opaque, and it’s not clear how to get the exact answers. At what point do the jobs that the data centers bring in materialize? How do you judge those things? Frequently, local councils don’t know how to answer these questions.
I think state governments need to create a regular process that someone who wants to build a data center has to go through to expose all of the information that the local communities need. Without that information, some residents will be like, “Why would you do this? You’re trying to do some shady backroom deal.” It’s usually not, but without that information, it’s hard to justify.
The other side of it is the permitting side. We have this long-standing problem in the telecommunications area where providers need to go in and get permissions in each locality they operate in to deploy their infrastructure. Energy suffers from a similar problem. Right now, in some areas, connecting a new data center to the grid can take up to seven years, which is why a lot of these places are starting to look at putting in their own on-site energy production.
So we have a very fragmented and inefficient way to get the energy into these data centers. We have a fragmented and inefficient way to build the necessary telecommunications networks. And then we have the legitimate problems at the local level.
What are you most optimistic about regarding AI or AI policy?
So Meta has these Ray-Ban glasses with the cameras and the AI that you can talk to, right? People get upset about them because they don’t want to have their picture taken in public, and we’re still working through the privacy issues there. But think about what this does for people who are visually impaired: they now have, on their face, something that helps them navigate the world. This person can now participate in society in a way that they couldn’t before.
It’s the same thing with AI. People of below-average intelligence do okay in life. They’re able to scrape by. Now they have a personal assistant who helps them understand the incredibly complicated modern world. So AI is obviously going to help us do even more impressive stuff at the frontier of human knowledge and capability. But it’s also going to allow more people to participate more fully in society. That can only be for the benefit of society, in my opinion.
