An economist on the impact of AI

[Bernard]

I'm glad to read this article from an economist, Paul Krugman. The negative side of technological progress seems to be always swept under the rug by people who study economics.

He touches on the subject, from the other thread here, of attention spans and reading. Subjects covered:

1. How mechanized agriculture made America less healthy
2. How modern manufacturing hollowed out cities
3. Contraception and women’s changing role
4. Smartphones and the rise of distraction
5. The social and psychological impacts of AI

I'm posting the whole thing here because I don't know if the link will work if you don't subscribe. But the link follows because he has some interesting charts to accompany the text.

https://open.substack.com/pub/paulkrugman/p/technology-and-social-change?r=n5ktx&utm_campaign=post-expanded-share&utm_medium=web

Technology and Social Change
Rising GDP isn’t the whole story
Paul Krugman

It’s sometimes hard to believe that ChatGPT was first released to the public less than four years ago. At this point AI is everywhere. This may be the most rapid adoption of a major new technology in history.

Despite AI’s ubiquity, its economic impact remains unclear. We don’t yet know what it will do to productivity, to employment, to wages or to income and wealth inequality. These are important issues, and I will be writing about them in the weeks ahead.

However, it’s important to realize that the ramifications of new technologies are much more than just productivity growth. They can indeed allow the economy to produce more goods with a given amount of resources. As I explained last week, “total factor productivity” is in fact the way economists measure the rate of technological progress. But new technologies also change society by altering the nature of work, where we live, how we interact with ourselves as well as others. Indeed, technological change can have profound social impacts even when its payoff in terms of higher GDP appears modest.

We know from history that the social changes caused by technological change aren’t always for the better. Sometimes technology makes society worse off in important ways. Sometimes it transforms society in ways some people find undesirable.

So this week I’m going to temporarily put the strictly economic impacts of technology aside and instead talk about technology and social change. As with last week’s primer, I will look at historical episodes as a way to gain insight into possible outcomes as AI diffuses through our society.

Beyond the paywall I will consider the following:

1. How mechanized agriculture made America less healthy

2. How modern manufacturing hollowed out cities

3. Contraception and women’s changing role

4. Smartphones and the rise of distraction

5. The social and psychological impacts of AI

Mechanized agriculture and American health

The United States began as a nation of farmers, with only a small fraction of the work force engaged in nonagricultural occupations and an even smaller fraction living in cities. Before 1840, agriculture’s share of the labor force barely declined. But it then began falling rapidly, so that by the late 19th century farmers were a distinct minority of the American population.

Simultaneous to the fall in agricultural workers as a share of the population, urban areas grew rapidly, especially larger cities. In 1840 there were only 3 cities in the United States with more than 100,000 people, accounting for only about 3 percent of the population. By 1900 there were 38 such cities, containing 19 percent of the population.

What caused this great rural-to-urban transformation? The most important cause was the introduction of agricultural machinery, like the horse-drawn mechanized McCormick reaper, patented in 1834. The introduction of these types of agricultural machines allowed a much smaller share of the labor force to feed the nation, leading to a shift to non-farming occupations and urban living.

This was not an entirely a good thing. Before the advent of modern sanitation, cities were often both socially chaotic and unhealthy. One little-known fact about 19th century America is that for much of that century, as urbanization progressed, Americans became shorter, a sign of deteriorating health and nutrition, even as GDP per capita rose:

While there is extensive debate about exactly what caused the deterioration in American health, the shift away from an overwhelmingly agricultural society must be a large part of the explanation. Thus history tells us that the social consequences of technological change may not be unambiguously positive even if it raises productivity.

Modern manufacturing, the hollowing out of cities, and urban disorder

Recently many people, including Azeem Azhar and myself, have been trying to make sense of the economics of AI by harking back to the transformation of factory design that took place once businesses learned to take full advantage of electrification. As a result of shifting from on-site generated power such as coal-fired boilers to electricity, the physical process of production changed radically. Squat multistory factories gave way to sprawling, single-story factories in which each machine had its own motor. And, in many cases, trucks, rather than railroad spurs, delivered materials and picked up products.

One important consequence of this change was that the need both for more land and for easy truck access led to a movement of manufacturing out of its traditional urban locations. We typically think of deindustrialization as a problem that began in the late 1970s with the rise of globalization. But, in fact, there was an earlier wave of deindustrialization within cities: in the 1950s, manufacturing employment in central cities began declining rapidly as industry became “suburbanized.”

While this move made business and economic sense, it had quite negative, even dire, social consequences. Social disorder in major U.S. cities increased drastically during the 1960s and 1970s, most visibly apparent in soaring crime rates:

What caused this social collapse? In his classic analysis “When Work Disappears” the great sociologist William Julius Wilson attributed much of the source of the problem to the emergence of mass inner-city joblessness, which itself was the result of the disappearance of urban blue-collar employment. Wilson’s analysis has held up very well over time.

However, if the social crisis of urban Blacks during the 60s and 70s was largely caused by the disappearance of urban blue-collar jobs, this in turn implies that this social crisis was to an important extent a consequence of technological change. Modern manufacturing technology precipitated an exodus of industrial jobs from America’s urban cores, but this exodus left many urban Blacks stranded.

By the way, the same thing happened to many rural Whites after around 1990, as the rise of a knowledge economy left much of the heartland similarly stranded.

As with the mechanization of agriculture, I’m not saying that we should somehow have prevented technological change. The point, rather, is that even when technologies make the economy more productive, they often lead to major social change — and this change can have dire consequences for some people.

The advent of oral contraception and women’s changing role

Some readers may not be accustomed to thinking of modern contraception as a technology, but of course it is. The invention of easy contraception represents a massive example of how focusing on the way technology affects GDP can miss its most important consequences for society.

Claudia Goldin received the 2023 Nobel Prize in economics “for having advanced our understanding of women’s labour market outcomes.” She has worked on many aspects of this subject, as well as other issues (her work on wage inequality transformed my own understanding of the subject), but perhaps her most influential research has been on the impact of oral contraceptives, aka The Pill. In a 2002 study co-authored with Lawrence Katz, Goldin showed that the pill — first made publicly available in 1960 — “did not diffuse among young single women until the late 1960s when a series of state law changes reduced the age of majority and extended ‘mature minor’ decisions.”

Availability of the pill helped sustain a trend toward rising female participation in the paid labor force that began in the mid-1950s. By increasing the total American supply of labor, the pill helped generate growth in GDP. Yet the social consequences were far more profound than the dollar gains.

As Goldin and Katz documented, widespread use of the pill quickly led to a transformation of women’s life and career choices:

The careers of women and their age at first marriage both changed significantly in the United States with cohorts born just prior to 1950. Women first began to enter professional programs, such as medicine and law, in large numbers in 1970 and their entry caused the fraction female in these programs to rise steeply. Women were 10 percent of first year law students in 1970, but were 36 percent in 1980. The fraction married among young college graduate women decreased for the same cohorts. Among the cohort of female college graduates born in 1950, almost 50 percent married before age 23, but fewer than 30 percent did for those born in 1957.

Why this rapid shift? They explain:

The pill directly and immediately lowered the costs to women of engaging in long-term career investments by giving them almost complete certainty and safety regarding the pregnancy consequences of sexual activity. The delay of marriage, beginning a year or two later, endowed the pill with a “social multiplier” or indirect effect by reducing the costs in the marriage market to women who delayed marriage to invest in careers. The relative increase of women to men in professional programs began its rapid ascent in 1970, just as the first pill cohorts began to graduate from college.

The pill, then, rapidly transformed America from a society in which very few women were able or willing to pursue professional careers to one in which many high-skill jobs are performed by women. We are still a long way from true gender equality, but we are far closer than we used to be.

Again, this shift surely increased GDP, not just by allowing more women to participate in the paid work force but also by allowing society to make much more use of half the population’s talents. But nobody believes that growth accounting is the most important consequence of the profound change in gender roles. The main point is that we became a very different society.

Did we become a better society? There is a significant movement in conservative political circles, most famously represented by the late Charlie Kirk, that says that these changes made society worse. They view the phenomenon of women pursuing careers while delaying marriage and childbirth as a bad outcome that should be reversed.

I disagree and find this movement appalling, but that’s an argument for another day. For now, the point is that contraception is a major example of a technology whose most important consequence has been social change — change that not everyone likes — rather than economic growth.

And there’s a more recent example of a socially transformative technology with surprisingly limited economic impact where the social consequences appear to be predominantly bad: smartphones.

Smartphones and the rise of distraction

The first iPhone was released in 2007. By 2013 a majority of U.S. adults owned smartphones; today the figure is 91%.

As I noted last week, the economic payoff to this rapid technological transition is surprisingly hard to find in the data. Total factor productivity, the standard economic measure of technological progress, has been much slower over the past 20 years than it was in the previous decade:

But never mind the economic data. Smartphones have drastically changed the way we relate to each other and to ourselves. And almost all the evidence shows that they have changed society for the worse. As a fact sheet from the Columbia University Department of Psychiatry — similar in its conclusions to what many studies say — puts it:

Smartphones have transformed the way we communicate, learn, and entertain ourselves. However, their omnipresence can lead to compulsive use and a sense of dependency. The constant stream of notifications and updates can create a sense of urgency and a fear of missing out, leading to increased anxiety and stress. Furthermore, the excessive use of smartphones can interfere with sleep, which is crucial for mental health.

Social media platforms, while enabling us to connect with others and share experiences, can also contribute to feelings of inadequacy and low self-esteem. The tendency to compare oneself with others and the desire for validation through likes and comments can lead to a distorted self-image and feelings of worthlessness.

Moreover, studies have shown a correlation between heavy social media use and depression, anxiety, loneliness, and suicidal ideation. The platforms are designed to be addictive, using algorithms that feed users content based on their preferences and interactions, keeping them engaged for longer periods.

These effects are worst for young people, especially girls, but they also apply to many of us. It’s true that smartphones can provide connection for some people who would otherwise be highly isolated, such as the elderly. They can be liberating in some cases. I personally really appreciate the ability to hold video conversations with family members when we’re temporarily on different continents. But overall, smartphones have clearly done a great deal of social harm.

It’s also increasingly clear that our addiction to the infinite scroll is doing long run intellectual harm. John Burn-Murdoch at the Financial Times documents a rapid decline in reading whose timing lines up precisely with the rise of smartphones;

He argues that we are becoming a “post-literate society,” and it’s hard to disagree.

Meanwhile, Noah Smith argues that smartphones, by reducing human connection, are a prime driver of declining fertility.

I could go on, but you get the point. Smartphones are a prime example of a technology whose most important impacts are social and psychological, rather than economic — and those social and psychological impacts are mostly bad.

There are now fitful moves to contain some of the damage, for example by banning smartphones in schools and attempting to ban children from social media. Better late than never.

Unfortunately, AI may have even worse social consequences than smartphones.

Social and psychological impacts of AI

My sense of direction is not what it used to be. I don’t think it’s advancing age, so far. I blame Google Maps: relying on my smartphone for directions has caused my ability to learn routes on my own to atrophy.

That’s a trivial example of a much bigger problem. I hear countless tales of educators dismayed to receive student essays that appear to have been partly or wholly written by ChatGPT. The issue isn’t simply the grading: A student who leaves research and writing up to a chatbot doesn’t learn the subject. That is, they will not be able to think through the topic independently.

Nor is this solely a problem for education. Workers can fail to learn necessary skills when they use AI to guide them in their tasks. Moreover, institutional memory and skills within a company may be altogether lost when workers are replaced by AI.

Reportedly this is already happening in the AI industry itself. According to CNN,

The vast majority – 71% – of AI-related job postings by S&P 500 companies on LinkedIn, like those for data analysts or machine learning engineers, are for senior-level positions … Big businesses are in bidding wars to land the same experienced pool of talent, raising questions about how they can attract the next generation of talent.

Then there are the psychological impacts of AI. Smartphones have already weakened and attenuated human connection. AI, we already know, can sometimes sever connection to others completely. There are now many well-known instances of people who develop unhealthy and sometimes fatal relationships with their chatbots. These aren’t typical, but they are surely extreme examples of the way AI can seduce us into self-destructive behavior. Furthermore, chatbot use among the young could potentially impede their ability to form healthy relationships in life. For example, a recent terrifying article documents how many boys, as young as 12 years old, are forming “romantic relationships” with chatbots.

I could go on, but I won’t pretend to be an expert on these social and psychological issues. Nor will I try today to offer solutions. My point for now is simply that the most profound and disturbing impacts of AI will probably involve its impacts on society rather than its effects on economic growth, or even on income and wealth inequality.

Nonetheless, these economic impacts are my usual turf, so I will return to them next week.

Check out my MasterClass on Economics

[Axtremus]

TL;DR: I guess AI will impact society bigly, but I don’t know how (but let me write about a few other things that impacted society bigly so it looks like I know something).

Hopefully he will indeed return to his usual turf and talk “economic impact.”

[Daniel]

Data centers are wrecking havoc on our natural resources, our democratic system of local self-government, our environment, and the health of our citizens.

Do you see what I did there? I wrote a post that required a. a brain, b. the ability to use it, and c. 11th grade English writing skills.

No. B.A., M.A., or Ph.D. required.

You're welcome. NNTTM.