The AI Economy (part 2)

Previous | Next

For this topic, my posts correspond to the three parts of Roger Bootle’s The AI Economy: Work, Wealth and Welfare in the Robot Age. Part 2 of his book is titled “Jobs, Leisure, and Incomes.”

Work and leisure

He begins with the impact of technology on the division of life between work and leisure. New technologies are often labor-saving devices, but the labor they save can either be channeled into other forms of work or enjoyed as leisure. Industrialization has had some of each effect, and Bootle expects this to be true of artificial intelligence and robotics too.

Industrial work has often been rather dreary, and many workers have felt that they are little but slaves to machinery. Not surprisingly, a shorter work week has been a traditional goal of organized labor. “Between 1870 and 1998, in the highly industrialized countries the number of hours worked per annum per employee fell from 2,950 to 1,500.” Divide that by 52 weeks, and it implies a reduction in the average work week from about 57 hours to 29. If 29 hours seems low, remember that it averages together full-time and part-time workers, many of whom might prefer either fewer or more hours. It may also seem low to Americans because our work hours are at the high end among industrialized countries, most of which have more vacation time than we do.

Although the long-term trend toward leisure is very real, it can be exaggerated. In the early twentieth century, John Maynard Keynes envisioned a reduction of the typical work week to 15 hours. Many social scientists have been surprised by the long hours many professionals work today.

Economists have looked for a relationship between hours worked and personal income, but the relationship is not simple. On the one hand, economists see an “income effect” in which some people see less need to work as their hourly income rises. On the other hand, economists see a “substitution effect” in which the substitution of leisure for work imposes the greatest opportunity cost on the highest-paid workers. They use this second effect to explain why so many well-paid workers seem so driven by their work.

Complicating the picture are all the non-financial reasons that people work: because they find the work intrinsically meaningful, enjoy winning in a competitive environment, enjoy the social relationships on the job, and so forth. Also, some long hours may be involuntary, imposed by employers as a condition of employment. Once upon a time, the wealthy were a more distinct leisure class, but today’s high-paid workers face more pressure to work, whether they like it or not. “Those wanting to work more are typically low-paid waiters or cleaners. Those wanting to work less are typically highly paid doctors or other professionals.”

In this complex situation, the effects of smarter machines are not likely to be simple. On the one hand, people who lack technical skills may have even more trouble finding as much work as they want. But people who are comfortable with new technologies may find the high-tech society liberating. They can turn many tedious tasks over to labor-saving devices, while enjoying the good pay and reduced hours that come with enhanced productivity. Many of the qualitative benefits of meaningful and creative work can also be achieved in unpaid activity, such as creative hobbies and volunteer work. As usual, Bootle emphasizes the positive potential of the AI economy rather than its dehumanizing possibilities. “Individuals themselves and society as a whole will be able—and indeed will need—to choose the right balance between work, rest, and play. And by ‘right’, I mean the balance that best suits them.”

Bootle reports that the countries with relatively low work hours, particularly northern European countries like Denmark, score higher on measures of personal happiness. A further reduction in work due to advanced technologies is not necessarily something to fear. That’s assuming that a lot of human work remains, and that people who want paid employment can get a decent piece of it.

Future jobs

Bootle complains that too many prognosticators lack the imagination to envision a world in which human activity is different but not necessarily less rewarding. They underestimate how much we humans will adapt to our changing technologies, as we always have. We will do less of some things but more of others, creating jobs that we can hardly imagine today. Who in an earlier phase of industrialization imagined the internet or the job of software engineer?

Bootle is highly skeptical of complete machine autonomy, believing that humans are much more likely to be assisted than replaced by technology. He discusses the possibility of self-driving vehicles at some length, concluding that computer-assisted driving is much more feasible. (I love my car’s smart cruise control that adjusts my speed to the car in front of me and keeps a safe distance.) Completely driverless cars have a lot of drawbacks: inability to anticipate every possible situation, very high cost, the threat of being taken over by hackers, and the deterioration of human driving skills so you don’t have them when and if you need them. The dangers of autonomous weapons systems are even more obvious. One of the manpower challenges in the military is that so-called “unmanned platforms” actually need to be monitored by technically qualified and sensible people.

The impact of artificial intelligence can be different from what people might expect based on past experience with automation. It is no longer the case that mental work is irreplaceable by machinery, while manual work is most vulnerable. Routine mental work like translating, interviewing or calculating is at risk, but “many skilled, manual jobs look safe for the foreseeable future. These include plumbers, electricians, gardeners, builders, and decorators.” A model for the worker of the future will be a creative person with the skill to work with a robot. “The losers in the AI economy will be those who are doing essentially robotic jobs and/or are themselves irreducibly robotic when dealing with other people.”

Although some jobs will no doubt disappear, Bootle sees many areas of increased worker demand—health care providers using artificial intelligence to diagnose conditions and robots to assist surgery; more technically assisted caregivers “as people get richer and older”; organizers of leisure to provide interesting and creative experiences off the job; human relationship experts; and of course, technicians to design and program the robots and help people learn to use them. He believes that “new jobs that barely exist today will spring up and multiply. Accordingly, I see no reason why the robot- and AI-infused economy of the future cannot be accompanied by full employment.”

Impact on inequality

Economic inequality within developed countries has been increasing, and the unequal distribution of technical skills is one reason why. Globalization aggravates the situation, since it puts workers with less education and skills in competition with similar workers in poorer countries. Some observers think that additional technological advances can only make things worse. Yuval Harari says, “In the coming century or so, humanity will divide into two classes of people—the gods and the useless.”

Bootle also discusses the potential for greater inequality between businesses. Producers of a popular digital product or service can dominate markets around the globe. This creates “winner-take-all” markets where millions of consumers can choose the global leader over local competitors. On the other hand, cutting-edge technologies can also be very disruptive. A startup company or new product can rise up to challenge a market leader, just as Apple challenged IBM in the 1980s. Entrepreneurs can rise from obscurity to fame in a few weeks.

Although people will need new skills to participate fully in the AI economy, opportunities to acquire skills should also abound. The situation reminds me of the need for mass literacy in the twentieth century. The response was a huge expansion of public education, facilitated by rising productivity in industry and the mass movement of children from farms and factories to schools. If new technologies create a productivity revolution in services, technical training services ought to be more widely available at modest cost. A new generation with early exposure to digital technologies may acquire useful skills pretty readily.

Technological advances have the potential to bring down the cost of services in general, just as assembly-line production brought down the cost of manufactured goods in an earlier era. If so, even the relatively worse off may be able to live better than they do today. As of now, the cost of higher education, legal services and health care remain stubbornly high, but the AI economy may find ways to make them more cost-effective.

Taking a moderate position, Bootle predicts neither a major increase nor a major reduction in inequality, but he does anticipate continued improvement in living standards. He says that “it is likely that the balance of the factors listed…will enable the mass of people to enjoy increasing incomes even as robots and AI start to proliferate. And it is perfectly plausible that there will not be a significant increase in income inequality.”


Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: