SA Forum is an invited essay from experts on topical issues in science and technology.

In Coningsby, a Benjamin Disraeli novel published in 1844, a character impressed with the technological spirit of the age remarks, “I see cities peopled with machines. Certainly Manchester is the most wonderful city of modern times.”

Today, of course, Manchester is mainly associated with urban decline. There is a simple economic explanation for this, and one that can help guide cities and nations as they prepare for another technological revolution.

Although new technologies have become available everywhere, only some cities have prospered as a result. As the late economist and historian David Landes famously noted, “Prosperity and success are their own worst enemies.” Prosperous places may indeed become self-satisfied and less interested in progress. But manufacturing cities such as Manchester and Detroit did not decline because of a slowdown in technology adoption. On the contrary, they consistently embraced new technologies and increased the efficiency and output of their industries. Yet they declined. Why?

The reason is that they failed to produce new employment opportunities to replace those that are being eroded by technological change. Instead of taking advantage of technological opportunities to create new occupations and industries, they adopted technologies to increase productivity by automating their factories and displacing labor.

The fate of manufacturing cities such as Manchester and Detroit illustrates an important point: long-run economic growth is not simply about increasing productivity or output—it is about incorporating technologies into new work. Having nearly filed for bankruptcy in 1975, New York City has become a prime case of how to adapt to technological change. Whereas average wages in Detroit were still slightly higher than in New York in 1977, they are now less than 60 percent of the latter’s incomes. At a time when Detroit successfully adopted computers and industrial robots to substitute for labor, New York adapted by creating new employment opportunities in professional services, computer programming and software engineering.

Long-run economic growth entails the eclipse of mature industries by new ones. To stave off stagnation, my own research with Thor Berger of Lund University suggests that cities need to manage the transition into new work (pdf).

Such technological resilience requires an understanding of the direction of technological change. Unfortunately, economic history does not necessarily provide obvious guidance for policy makers who want to predict how technological progress will reshape labor markets in the future. For example, although the industrial revolution created the modern middle class, the computer revolution has arguably caused its decline.

To understand how technology will alter the nature of work in the years ahead, we need to look at the tasks computers are and will be able to perform. Whereas computerization historically has been confined to routine tasks involving explicit rule-based activities, it is now spreading to domains commonly defined as nonroutine. In particular, sophisticated algorithms are becoming increasingly good at pattern recognition, and are rapidly entering domains long confined to labor. What this means is that a wide range of occupations in transportation and logistics, administration, services and sales will become increasingly vulnerable to automation in the coming decades. Worse, research suggests that the next generation of big data–driven computers will mainly substitute for low-income, low-skill jobs over the next decades, exacerbating already growing wage inequality (pdf).

If jobs for low-skill workers disappear, those workers will need to find to jobs that are not susceptible to computerization. Such work will likely require a higher degree of human social intelligence and creativity—domains where labor will hold a comparative advantage, despite the diffusion of big data–driven technologies.

The reason why Bloom Energy, Tesla Motors, eBay and Facebook all recently emerged in (or moved to) Silicon Valley is straightforward: the presence of adaptable skilled workers that are willing to relocate to the companies with the most promising innovations. Importantly, local universities, such as Stanford and U.C. Berkeley, have incubated ideas, educated workers and fostered technologies breakthroughs for decades. Since Frederick Terman, the dean of Stanford’s School of Engineering, encouraged two of his students, William Hewlett and David Packard, to found Hewlett–Packard in 1938, Stanford alumni have created 39,900 companies and about 5.4 million jobs.

For cities to prosper, they need to promote investment in relevant skills to attract new industries and enable workers to shift into new occupations. Big-data architects, cloud services specialists, iOS developers, digital marketing specialists and data scientists provide examples of occupations that barely existed only five years ago, resulting from recent technological progress. According to our estimates, people working in digital industries are on average much better educated and for any given level of education they are more likely to have a science, technology, engineering or mathematics (STEM) degree. By contrast, workers with professional degrees are seen less often in new industries, reflecting the fact that new work requires adaptable cognitive abilities rather than job-specific skills. It is thus not surprising that we find San Jose, Santa Fe, San Francisco and Washington, D.C., among the places that have most successfully adapted to the digital revolution.

The cities that invest in the creation of an adaptable labor force will remain resilient to technological change. Policies to promote technological resilience thus need to focus on the supply of technically skilled individuals and encouraging entrepreneurial risk-taking. For example, the National Science Foundation recently provided a grant to North Carolina Central University to integrate entrepreneurship into scientific education. More such initiatives are needed. Furthermore, immigration policies need to be made attractive to high-skill workers and entrepreneurs. Cities like New York and London owe much of their technological dynamism to their ability to attract talent.

Meanwhile, it is important to bear in mind that policies designed to support output created by old work are not a recipe for prosperity. Whereas General Motors has rebounded since its 2009 bailout, Detroit filed for Chapter 9 bankruptcy in 2013. Instead of propping up old industries, officials should focus on managing the transition of the workforce into new work. The places that do so successfully will be at the frontier. As argued by Jane Jacobs in more colorful terms: “Our remote ancestors did not expand their economies much by simply doing more of what they had already been doing…. They expanded their economies by adding new kinds of work. So do we.”