Welcome back to the Economics of AI. This week's module is on AI and economic growth, and I will talk about how we can think of the effects of advanced artificial intelligence on the process of growth. I will intentionally take a long view of the issue. I will go far back in history and also try to look forward into the future and see how economic growth has changed throughout the different epochs of humanity. The history of mankind is a history of different ages. In this week's module, I will follow that structure. It all started out with the Stone Age, at first the Paleolithic. Then there was the Neolithic revolution when people started to settle down when they engaged in farming and domesticated animals. Throughout these times, Malthusian forces ruled the world. The human population was determined by the same forces as all species of animals. We multiplied when our production of food increased, and the population declined when food production fell. Human standards of living were essentially unchanged for millennia. Even the Neolithic Revolution did not change that. It made it easier to produce food but it just increased the population and did not improve living standards, so GDP per capita was essentially unchanged. There are even historians who emphasize that living conditions of humans may have deteriorated when we settled down because we became much more vulnerable to disease. All this suddenly changed with the advent of the Industrial Revolution in the late 17th century, the dawn of the Industrial Age. This represented the most fundamental change in the process driving human living standards so far. Growth rates suddenly increased by an order of magnitude. Before, GDP per capita was essentially constant. Since the Industrial Revolution, at first the advanced countries and then a growing number of developing countries, have seen sustained increases in living standards as measured by GDP per capita. In our second part of this module, we will therefore go over models that explain the forces that have driven economic growth since the Industrial Revolution. We will review the classic Solow model and related models of neoclassical growth. For many of you, this material may be familiar. If you already are an expert in Solow growth, feel free to just skim that part of the material. Finally, we will analyze what to expect in the age of AI. One of the points that I will try to emphasize is that the range of possibilities, the range of potential outcomes from advanced AI is really fast. When teaching economic growth at the undergraduate level, macroeconomists like to show graphs of advanced countries during the 20th century, which suggests that economic growth is this really predictable force that always hovers around something like two percent. But if we zoom out a little bit, this looks very much like a fluke, a brief episode in the history of humanity, rather than a fundamental economic law. Now, we are at the birth of the age of AI. It's really difficult to predict what this will imply for humanity, but what we'll see is that our economic models of growth allow for a pretty wide range of possibilities that all are compelling at some level. And given the revolutionary nature of AI, we really have to seriously consider this vast range of possibilities. If some of the predictions of technologists turn out to be true, we may see a similar change in our growth trajectory as what we saw when the economy went from stagnating living standards in the Malthusian age to the exponential growth in living standards that we have seen since the Industrial Revolution. Let's start with a brief look at the data. Unfortunately, the data does not go back 50,000 years when mankind first developed language, but it goes back to the Middle Ages. And the country for which we have the most reliable data series is England. On this graph, you can see living standards in England from the beginning of when we had data 1246 until now. The upper graph and the lower graph show the same data but the upper graph is in a logarithmic scale, and the lower graph is on a linear scale. From the beginning until the Industrial Revolution, and this is most visible on the logarithmic scale, output per capita fluctuated around a fixed level of something like a thousand US dollars per capita per year. It was only at the beginning of the Industrial Revolution that we see that the curve started to bend up, and how much it rose is most salient under linear scale in the bottom graph, which shows the typical signs of exponential growth since the Industrial Revolution started. The first lesson of the segment will focus on a model of the Malthusian economy that explains why living standards were stagnating for so long. So this data was from Britain, the country that first went through the Industrial Revolution, and it more or less corresponds to what the Solow model will tell us in one of the next lessons, but let me also show you a graph that depicts growth data for the world as a whole based on our best estimates. This graph depicts average growth rates for all the countries for which data was available in a given decade averaged over 10-year periods to reduce noise. What we can see on this graph is that for the world as a whole, it looks more like growth rates have been rising exponentially since the Industrial Revolution. If this exponential growth trend continues, we may need rather different growth models going forward than our traditional growth models of the Industrial Age in which growth rates were constant. And we will look at some of those models in the lesson on growth and AI. There's also one mandatory reading that is part of this week's materials, a survey paper on economic growth and a transformative AI authored together with Phil Trammell. You will find it right after this intro video. Please go through it carefully as it covers some of the materials in much more detail than what I could cover in this video.
