Could be the key word here, because it’s still too early to know how this will play out. However, we have places to turn for clues. After all, this is not the first time that tasks performed by humans have been taken over by machines.
“One way to think about artificial intelligence is to continue the process of automation, but to target mental or cognitive tasks that people do as opposed to physical tasks,” says Benjamin Jones, professor of strategy at Kellogg.
Drawing of two of his own recent studies on automation, Jones lays out a few different scenarios for how AI could transform the economy. He points to two specific factors that will likely determine how much of an impact AI ultimately has—and how the average worker fares.
Factor 1: Whether AI will target bottlenecks in the economy
The first of these factors is where in the economy, progress happens. If AI addresses sectors that are currently experiencing productivity problems, then its impact on the economy will be much greater than if it does not.
Compare, for example, historical advances in agriculture, where there are fewer bottlenecks, with those in IT, where there are many.
Hundreds of years ago, “you and I would both be farmers, almost certainly, because almost everyone was a farmer,” says Jones. Despite the much hard work that went into preparing the land, cutting down the trees, preparing the furrows for planting, watering, fertilizing and harvesting, most farms produced enough to feed the families who worked them, with perhaps a little sold in the market if they were lucky.
In advanced economies today, agriculture is highly automated — and highly productive. There are machines that level the land and water, machines that plant seeds in perfect rows, and machines that apply the exact amount of fertilizer needed. A single modern combine harvester can process over a hundred acres a day. “We’re producing more food than ever before, but almost nobody is a farmer” because of modern automation, Jones says. Recent estimates show, for example, that only approx 1.3 percent of American workers are employed on a farm.
In some ways, these machines were incredibly destructive – but because modern agriculture is so much more productive, eating has become much easier for everyone. “And then what happened was we freed up labor so it didn’t have to be farmers, so people could migrate to other jobs,” he says. “There is a process of turning the economy around, destroying many jobs, creating new kinds of jobs and people being redeployed into new roles in a more productive economy, in which people are on average significantly richer and live longer.”
Now fast forward to another technological advance: computers. Before IBM introduced its mainframe computer in 1952, human “computers” did the demanding work of number crunching. Since then, of course, the machines have taken over, to spectacular effect. Moore’s Law, which states that the number of transistors on a microchip roughly doubles every two years or so, has proven true for nearly 80 years and has led to life-changing innovations from the Internet to smartphones and applications such as Zoom. “You can have a video call in the back of a taxi with a family member 6,000 miles away. That would have sounded like fantasy to people in 1980,” says Jones.
But since Moore made his famous prediction, something strange has happened: “Productivity growth has been unusually slow,” says Jones. Despite impressive advances in computing power, the economy has not become much more productive, nor has the standard of living risen dramatically.
To understand this contradiction, Jones says, “you have to realize something very important about the way economies develop: what really matters is what we’re bad at.” That is, productivity does not depend on how efficient the economy is at its best, but at its worst.
Many tasks in agriculture are straightforward and highly repetitive — meaning they are easy to automate. This meant that the industry was able to dramatically increase its productivity without facing too many bottlenecks: its slowest, “worst” processes were still very fast.
On the other hand, while computing power has grown exponentially, it is often in the service of more cognitive, tailored tasks such as legal services. These have historically been more difficult to automate, leaving many bottlenecks. After all, it doesn’t matter how fast your computer can run if every output has to be written or checked manually.
And it’s not just professional services that are congested. Consider the time and effort it takes to travel across the country, cook a meal or visit your doctor, or build a power plant. These things haven’t really changed that much in fifty years because, for some reason, there are bottlenecks.
But bottlenecks don’t just slow growth in a given sector: they have a major effect on the entire economy. As a job is automated and becomes more productive, it shrinks in its share of the economy. That’s because the exit is becoming ubiquitous — and cheap. (Consider that agricultural production on all of America’s farms is now just 0.7% of GDP.) Meanwhile, it is the less productive tasks—the bottlenecks—whose share of the economy grows over time. This means that, in the end, most of the economy is devoted to sectors that do not increase productivity at all. This is where most advanced economies are today.
How much AI will transform our economy, then, depends on the extent to which it can increase productivity in those unproductive and expensive parts of the economy, such as healthcare, education, hospitality, transportation or electricity. Autonomous vehicles or robots that could do household chores, for example, could dramatically free up human labor, just as advances in agricultural machinery have done for farmers. Likewise, using AI to make individual doctors or educators significantly more productive could also help alleviate these bottlenecks.
For this kind of automation to truly change the game, AI “can’t just be the next smartphone,” says Jones. Ultimately, we got the smartphone, which improved some cognitive tasks like navigating a city or retrieving information, but didn’t quite drive productivity. Instead, AI “must go beyond that if it’s going to really bend the productivity curve.”
There is some reason to believe that this is possible. “In the sense that AI is a cognitive-oriented technology that targets many services, I think it targets the mass of the economy where the bottlenecks are,” he says.
Factor 2: Whether the AI will perform cognitive tasks much better than humans
The second factor that will determine the impact of AI on the economy is whether AI will replace human labor by having just barely better than us or dramatically better.
If AI is good enough to replace humans, but not much better than that (a scenario that is, at least immediately, plausible), that’s very bad news for human work.
Consider an automated checkout kiosk at a grocery store or a bot that provides customer service to airline passengers. Few would argue that these technologies are far superior to their human counterparts. Often, they are little worse than cashiers or agents. But they is cheaper and businesses are likely to implement the solution at the lowest cost that clients and customers will tolerate. In this scenario—where AI replaces human labor without dramatically increasing productivity—labor takes a lower share of income and we don’t see dramatic gains in living standards.
This leaves many workers worse off while providing little benefit to the overall economy.
But there is another, more optimistic scenario here. If AI does indeed prove transformative—for example, enabling a single radiologist to do the work of 15 radiologists, and a single coder to do the work of 15 coders, and so on—then we can expect a burst of economic growth that will enable everyone us to enjoy a higher standard of living (even if some radiologists and coders have to find other jobs). This will be true even if machines take over the vast majority of the work, if they exist at least some tasks that require human labor.
To understand why, remember how the share of the economy devoted to bottlenecks will always continue to grow, while tasks that can be easily automated will shrink. This means that as automation takes over more tasks, the remaining non-automated tasks will increase in importance and people will be better compensated for performing them. And in this “explosive growth” scenario, the economy will expand so quickly that these tasks will pay off big indeed.
Consider that today’s cellists are no more prolific than 17th century cellists. They play the same music, even the same instruments, and the demand for their work has only diminished as other forms of entertainment have appeared. “So why are they being paid 20 times more in real terms?” says Jones. The answer is that “in a sense, the cellist is the bottleneck.” That is, the productivity of the cellist has not increased, but because so many other parts of the economy I have increased (and thus the relative cost of so many other things has decreased), the cellist enjoys a higher standard of living.
That’s why Jones hopes for a “world where humans do a very small fraction of the work, but they do something, and computers become infinitely good at everything else,” Jones says. “If we become super productive with machines, it’s like we’re all cellists.”
