Making AI Serve Us All

I had this curiosity. Then I had parents who encouraged and appreciated the curiosity. Even though they didn’t understand all the things I was investigating or curious about at any point in time, they, whenever they could, tried to help. So, I think I was extraordinarily lucky to have that. We should never take that sort of support for granted. I think it’s just really scary sometimes, taking these big risks you have to take, to figure out whether or not this thing you’re interested in is interesting to other people or that it’s going to be useful, or that there’s an opportunity there. I’m just eternally grateful to all of the people in my family who were there to support all of what, to them, must have seemed silly and risky and crazy things that I wanted to do.

Moret: One of the themes in your book has to do with this widespread fear that many have in the United States of artificial intelligence killing jobs. Various studies have cited a certain percentage of jobs could go away within 10 or 20 years as a result of AI. You have a more hopeful outlook, I think, on that potential impact. Can you elaborate on how you think AI could evolve in a way that could actually enable job growth, innovation, and creativity in the United States?

Scott: Almost any technology we’ve ever invented that ends up having major impact and becoming ubiquitous in our lives starts off seeming a little bit scary. In many cases, there’s specific anxiety around technologies causing disruptions in jobs. My favorite example is the steam engine, which, in the late 18th century, was the first real mechanical substitute for human labor. When steam engines were deployed in factories and became part of the means of production, they radically transformed the economy. It did a ton of good, but it was disruptive for jobs.

In the early days of the technology, the benefits accrued to society because they had more ubiquitous, cheap manufactured goods than they had before. The benefits also accrued to folks with capital and folks with expertise. So, if you had enough money, you could invest in these expensive steam engines and build business concerns around them. If you were an expert, you could earn a really good living and create profitable businesses for yourself by designing and operating these machines as a means of production.

We’re constantly, as a society, faced with these problems that, on the surface, look like zero-sum games. Some of the challenges today in health care and climate change look very, very challenging to solve. But if you can turn a zero-sum game into a non-zero-sum one — where you don’t have the same constraints, where you can also think about expanding resources, creating abundance that didn’t exist before, relaxing constraints — then you can have a lot more flexibility in how you solve problems.

I think what we’re already seeing is some of those predictions made a few years ago that were very dire in the sense that so many jobs were going to get disrupted so quickly, were not as accurate as we thought they were going to be. A huge number of people will be able to pick up these tools and to use them to create new businesses, which will have benefits not just for the business creators, but for folks in their communities, employees, and the people who are benefiting from the things they create.

One of the things I want to be able to do in the role I’m playing right now is to try to make sure that these sophisticated technologies are packaged as platforms where other people can use them. I don’t think we’re going to get the full benefit from them if just a small number of people in very big tech companies that sit in coastal innovation centers are the ones who must have all of the imagination for what good the technology can do.

Moret: One of the stories you told in the book that really stuck with me started with your great-grandmother, who was born in 1898 and lived to 1997. In her life, she saw three big technological changes: ubiquitous electricity, ubiquitous refrigeration, and television. You use those to talk about what it means to be a platform technology with the potential to generate profound economic transformations across multiple industry sectors. Can you elaborate on what you mean by that term, as well as why AI fits in that category?

Scott: Electricity is a platform technology that has these self-reinforcing feedback loops. The technologies you build on top of electricity improve our ability to produce more energy. After the development of electric power distribution and its ubiquitous availability, we eventually invented digital computing. Digital computing as a platform lets us do things like design solar panels and better electric power generation facilities. It allows us to manage the loads on the electric power grid to better match consumption to production, or production to consumption. The availability of technology makes the technology itself better.

I would encourage people to not think about AI as one thing. It’s not the Terminator from the Arnold Schwarzenegger movies. AI is really a very broad collection of technologies that are being used already to power things in every part of our lives. It has this self-reinforcing feedback loop in that the more we use AI, the better AI becomes, because AI benefits from having large amounts of data and large numbers of interactions. I think AI is going to be the 21st century’s equivalent of electricity. It will be the most important platform technology we invent this century.

Moret: To me, the third big focus of the book is the potential for AI to be a key tool in enabling revitalization and economic opportunity in rural America. Could you lay out what you see as the key foundational elements of the rural growth agenda — for not just rural Virginia, but for America as well, and where AI could play into that?

Scott: I’ll start by telling a story to illustrate how accessible AI technology has become. Prior to joining Google, I wasn’t a machine learning expert at all. When I got to Google, I had the opportunity to work on this project in the ads technologies systems that required me to do a whole bunch of machine learning. When I started work on this project, I spent a couple of months reading a bunch of very complicated, daunting, graduate-level textbooks on statistical machine learning, going through stacks of papers, and coming up a very steep learning curve. Then I spent six months writing a whole bunch of super-complicated code to solve this problem.

We fast-forward 16 years to today. This thing I did — because of open-source software and cloud computing platforms and all the knowledge and training tools that are available to anyone, for free, online — I think a motivated high school student could have done that entire project in a couple of days. The more accessible these tools are, the more people in any community anywhere are going to be able to pick them up and solve problems that only they can imagine.

Moret: When you talk about rural development, you do highlight a few key programmatic and policy things that need to happen to enable this potential to be fulfilled. Can you touch on those?

Scott: When venture capital goes hunting for good businesses and good entrepreneurs in other parts of the country, there are good returns to be had. I think there’s a role for government to play in incenting some of that investment just so that we can get momentum and investment into those communities, after which we’ll absolutely see the returns and the benefits.

In order for these communities to be able to connect with their digital future, they actually have to be able to connect to the internet. You can’t run a modern, technology-heavy business when your business and your employees don’t have great broadband connectivity.

There are also some education things we have to do, partly about just making sure that role models exist for kids. The three businesses that were big in my community when I was growing up were tobacco farming, furniture manufacturing, and textiles. When you looked around and saw who had good jobs and who had opportunities, they were the people who had work in those three industries. There’s no reason whatsoever that a rural kid can’t have a beautiful career in technology and, increasingly, a beautiful career in technology where they don’t have to leave their community and their family. But we need to do a better job of showing kids how.

Moret: One of the things we’ve been excited about in Virginia is that Microsoft has been a leader here in rural development. We’ve had a very long and positive relationship with Microsoft in the Commonwealth. I wonder if you could talk a bit about Microsoft’s presence in Southern Virginia.

Scott: It’s one of the most sophisticated pieces of technology anywhere on the planet. We’ve created hundreds of very high-skill jobs in the Boydton Data Center. One of the interesting things I think you see is that these high-skill, high-wage jobs have a network effect inside the communities they’re in. They produce more jobs than just the jobs themselves.

There’s this really great book I would encourage everyone to read, by a Berkeley economist, Enrico Moretti, called “The New Geography of Jobs.” He describes this phenomenon where an engineering job in this data center is probably going to produce three, four, or five other jobs, just in terms of the economic halo effect it produces as you have these high-wage earners in these communities.

A small community like Boydton can have a big job creator, like the data center, which then results in this gravitational field that attracts technical people into the area, which results in more auxiliary businesses that have their own jobs. It’s easier now than it ever has been to create businesses that can serve global customers anywhere.

Moret: In your book, you said, “Whether it’s health care or a combination of health care and other grand challenges, we already possess the mechanisms to fund an AI Apollo program.” Can you elaborate on what your vision is there, what it would cost, and what you think the potential outcomes would be for economic competitiveness and growth and opportunity in the U.S.?

Scott: The reason I mentioned the Apollo program is that we didn’t need to go to the moon. The beautiful thing about the Apollo program is that it created this single galvanizing vision for what human innovation could accomplish and this big, hairy, audacious goal that we could collectively go after. The goal itself is a little bit arbitrary, but the things we needed to do to accomplish the goal, the mission, were very, very strategic. Basically, our modern aerospace industry was born out of the space race, the Apollo program, and the investments we made in going to the moon.

We actually could do something even better than the Apollo program right now, because we have a bunch of very important problems that we could go solve. My favorite one is health care. Climate change is another interesting one where, in order to turn this from a zero-sum problem, we’re going to need to invent a whole bunch of technology that can make our carbon emissions much lower, as well as a whole bunch of new technologies we have to develop to get carbon out of the atmosphere. AI is a very, very important component in solving those problems, as well.

So, you can pick your moonshot here. In going after either of those moonshots, you solve a big, important problem for society, but you also produce an entire ecosystem that’s probably going to produce even more economic benefits.

Moret: Kevin, what a pleasure this has been. I just want to end by saying thank you so much for joining us today. Thank you for all you’ve done to advance technology and AI, and what you’re going to do in the future. Thank you especially for all you’ve done to support development in the Commonwealth of Virginia. We’re really proud that you’re from here, and hope we’ll be able to maintain a close relationship in the years ahead.

Scott: Thank you for what you’re doing, as well. I still consider Virginia my home. I think it’s one of the most beautiful and special places on the planet. So, I’m glad to know that you’re there working so hard on creating opportunity for Virginians, and for the Commonwealth, for a great future.