The future of work

Authors: Jonathan Grudin
Posted: Tue, March 24, 2015 - 12:23:12

Some researchers and pundits predict that automation will bring widespread unemployment. This is unlikely. The shift of some labor to technology has been in progress for decades, but in the past 5 years the United States added almost 12 million jobs. Where is the automation effect? What will materialize to shift us from fast forward to permanent reverse gear? What drives this fear?

In an earlier post, I mentioned an invitation to a debate on this topic after being among the optimists in a Pew Research Center survey. Pew’s respondents were divided. 48% believe technology will increase unemployment; 52% believe employment will increase. I was quoted: “When the world population was a few hundred million people there were hundreds of millions of jobs. Although there have always been unemployed people, when we reached a few billion people there were billions of jobs. There is no shortage of things that need to be done and that will not change.”

Four principal speakers at the Churchill Club forum on Technology and the Future of Work were eminent economists, including a former chair of the White House Council of Economic Advisors and a former director of the White House National Economic Council. The other four were technologists. A Singularity University representative insisted that within five years, all work would be done by intelligent machines. Jobs in China, he said, would be the first to go. The President of SRI said that it would take 15 years for all of us to be out of work. A third technologist exclaimed that the impact of technology now was like nothing he had ever seen.

“That’s because you didn’t live in the 19th century,” an economist said dryly.

Off-balance, the technologist responded, “Neither did you!”

“No, but I’ve read about it.”

The best guess at tomorrow’s weather: the same as today’s

Technologies eradicated occupations, yet the workforce grew. Americans employed in agriculture fell from 75% to about 2% in a little over a century—and that was after the industrial revolution transformed the Western world. Hundreds of thousands of telephone operators were replaced by technology in the late 20th century. The economists at the forum did not anticipate imminent doom, but some expressed reservations about the nature of the jobs that will be available and concern over growing disparities in income and wealth. Will workers who lose jobs retool as rapidly as in the past? We don’t know. Shifting from farming to manufacturing required major changes in behavior and family organization.

At one time, positively rosy views of automation-induced leisure were common. Buckminster Fuller predicted that one person in ten thousand would be able to produce enough to support ten thousand people. Since one person in ten thousand would want to work, he surmised, no one would have to work unless they wanted to. Arthur Clarke, another writer, inventor, and futurist, had similar views.

Alas, the evidence suggests otherwise. Hunter-gatherer societies were relatively egalitarian as they struggled to survive. Agricultural self-sufficiency and the ability to satisfy basic necessities did not lead to leisure—hierarchies of privilege sprang up and funneled most resources to the top 1%. Growing income inequality in the United States may be a regression to the norm [1]. The 1% profit by finding ways to keep the 99% producing: There are armies to equip and pyramids to build.

Let’s bet against a job-loss pandemic. Despite the current enthusiasm for deep machine learning, the Singularity isn’t imminent. We hear more complaints about systemic malfunctions than breathless reports of technological adroitness. Deep Blue was an impressive one-trick pony focused on a constrained problem: a handful of objects governed by a limited number of rules. It was dismantled. The Watson search engine retrieves facts, which is lovely, but it’s our job to use facts. Four years after the Jeopardy victory, investors are pessimistic about IBM.

In the 1970s, it was widely predicted that programming jobs such as mine at the time would be automated away. This was music to management’s ears—programmers were notoriously difficult to manage and insisted on high wages, which they tended not to use for haircuts, suits, and country club memberships. Tools improved and programmers became software engineers, but employment rose.

The web brought more predictions of job loss—who would need software developers when anyone could create a web page? But markets appeared for myriad new products—developers didn’t go away— as did hundreds of millions of web sites. Anyone can create a site, but it requires an investment of time to learn a skill that will be exercised infrequently and needs to be maintained. It is more cost-effective to hire someone with a strong sense of design who can do a sophisticated job quickly. The new profession of web site designer flourished.

I visited my niece, a prosperous organic farmer. She has solar panels that supply a battery that maintains an electric fence. In the distance, solar panels power a neighbor’s well pump that irrigates a field. Solar already employs twice as many Americans as coal mining. Employment created by technology, and it is early days—the Internet of Things will bring jobs we can’t imagine.

Work that is not inherently technical benefits from technology. Almost any skill that could be considered as a career—cooking fish, growing orchids, professional shopping, teaching tennis—can be acquired more rapidly through resources on the web. Reaching a level of proficiency for which people will pay takes less time than in the past.

When I was 18, I knew the one “right way” to hit each tennis stroke and was hired to teach in a city park system. Twenty years later, I attended a weekend morale event at a tennis camp. There was no longer one right way to teach. Technology had changed coaching. My strokes were videotaped. The instructor identified all weaknesses. (Today, fifteen years later, machine vision might identify the problems.) The coach’s task shifted: She decides which of the five things I’m doing wrong to focus on first. She gauges how many I’m capable of taking on at once. A coach sizes up your personality and potential. Should she say “good job!” to avoid discouraging you or “try again, hold it more level!” to keep you going? A good instructor knows more about strokes and also understands motivation. Jobs are there. A weekend warrior could go online, have a friend videotape strokes, and perhaps find analytic software. How many will bother?

We all have agents

An employee of the science fiction author Robert Heinlein told me that Heinlein was methodical. He and his wife researched the planet to find the perfect place to live and settled on a plot between Santa Cruz and San Francisco. The only drawback was seismic. The Heinleins designed a house to “ride earthquakes like a ship rides the sea,” anchoring heavy furniture deep in the foundation.

When Heinlein worked on a book, he stayed in his room and gained weight. To avoid obesity he lost weight before starting a book, ending up back at normal. To the dismay of his agent, Heinlein was not that fond of writing and his royalties covered his needs, my friend said. Heinlein’s agent determined that his job was to find and entice the couple with expensive consumer items, to convince Heinlein to undertake the ordeal of writing another book.

We all have agents who benefit from our labor by convincing us to work for things we may not need. One can have reservations about consumerism, but the ingenuity to devise and market objects is a notable human skill.

Good jobs

“Will the new jobs be good jobs?” What does this mean? Were hunting, gathering, and farming good jobs? Working on an assembly line or as a desk clerk? Is an assistant professor’s six-year ordeal a good job? “Good” generally means high-wage; wages are set more by political and economic forces than by automation. The inflation-adjusted minimum wage in the U.S. peaked in 1968, almost 50% higher than today—wage growth will raise the perceived quality of many jobs. Globalization and competition can drive down wages and are enabled by technology, but are not consequences of automation. 

Despite the rapidly falling U.S. unemployment rate, there is uneasiness. Youth unemployment is high, not all of the new jobs are full-time, wage growth is rising more slowly than anticipated, income disparity is increasing. And it is reasonable to ask, “Could another shockingly sudden severe recession appear?”

I see students in high school and college who have none of the optimism my cohort did that good jobs will follow graduation. Anxiety is high, fed by reports they won’t live as well as their parents. Some of this is the reduction in support for education: Student debt was rare in my era. Because my friends and I were confident in our future prospects, we could take a variety of classes and explore things that were interesting, whether or not they had obvious vocational impact. We spent time talking and thinking. Well, and partying. Maybe mostly partying. Anyway, students I encounter today express a greater need to focus narrowly on acquiring job-related skills. There is still some partying. Maybe not as much.


Growth in U.S. employment is the strongest since the Internet bubble years. Productivity growth has tapered off—the opposite of what would be expected were automation kicking in. With little objective evidence of automation-induced joblessness, it is natural to wonder who is served by the discourse around employment insecurity.

Two beneficiaries come to mind. The top of the hierarchy benefits if the rest of us worry about unemployment and accept lower wages. Employers benefit when graduates are insecure. Also, technologists benefit in different ways. The reality may be that productivity gains from computerization have leveled off, but the perception that technology is having a big impact could be to the psychological and economic benefit of technologists. Who would want to lay off those writing code so powerful it will put everyone else out of work?


1. In the conclusion to the formidable 1491, Charles Mann suggests how the young United States resisted this norm.

Thanks to John King for discussions on these issues.

Posted in: on Tue, March 24, 2015 - 12:23:12

Jonathan Grudin

Jonathan Grudin is a principal design researcher at Microsoft.
View All Jonathan Grudin's Posts

Post Comment

No Comments Found