Are There Robot-proof Jobs?

When I give talks about AI and robots, they are often followed by a bit of Q&A. By far, the number one question I am asked from the audience is a variant of, “What should my kids be studying today to make sure that they are employable in the future?” As a dad with four kids under 20, I too have pondered this question at length.

If possibility one is true—that is, if robots take all the jobs—then the prediction of the author Warren G. Bennis will also have come true, that “the factory of the future will have only two employees, a man and a dog. The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment.” In other words, there would be no robot-proof jobs.

But if possibility two or possibility three comes to pass, then there will be robot-proof jobs. What will they be? A good method for evaluating any job’s likelihood of being automated is what I call the “training manual test.” Think about a set of instructions needed to do your job, right down to the most specific part. How long is that document? Think about a posthole digger versus an electrician. The longer the instruction manual, the more situations, special cases, and exceptions exist that need to be explained. Interestingly, when surveyed, people overwhelmingly believe that automation will destroy a large number of jobs, but also overwhelmingly believe that their own job is robot-proof. In other words, most people think that the manual to do their job is large while other people’s job manuals are smaller.

The reason the training manual test works is because writing a manual on how to do a job is a bit like programming a computer or robot to do a job. In a program, every step, every contingency, every exception, needs to be thought through and handled.

One wonders if there are there some jobs that can’t be written down. Could anyone write a set of instructions to compose a sonata or write a great novel? How you answered our big foundational questions probably determines what you think on this question. To those who think they are machines, who are monists, there is nothing mysterious about creativity that would keep machines from mastering it, whereas those on the other side of that gulf see creativity as a special, uniquely human ability.

Below are several groups of jobs that, regardless of your beliefs about the capabilities of robots, should be stable for a long time.

Want to see how robot-proof your job may be? Take our interactive quiz!

Jobs Robots Can Do but Probably Never Will

Some jobs are quite secure and are accessible to a huge range of the population, regardless of intellect, educational attainment, or financial resources, because although a robot could do them, it doesn’t make economic sense for them to do so. Think of all of the jobs people will need for the next hundred years, but only very occasionally.

I live in a home built in the 1800s that contains several fireplaces. I wanted to be able to use them without constantly wondering if I was going to burn the house down, so I called in “the guy” for old fireplace restoration. He took one look at them and started spouting off how they clearly hadn’t been rebuilt in the nineteen-sometime when some report came out in England that specified blah-blah-blah better heat reflection blah-blah-blah. Then he talked about a dozen other things relating to fireplaces that I tuned out because clearly this man knew more about fireplaces than anyone else I would ever meet, or he was a convincing enough pathological liar that I would never figure him out. Either way, the result is the same: I hired him to make my fireplaces safe. He is my poster child of a guy who isn’t going to be replaced by a robot for a long time. His grandkids can probably retire from that business.

There are many of these jobs: repairing antique clocks, leveling pier-and-beam houses, and restoring vintage guitars, just to name a few. Just make sure the object you’re working on isn’t likely to vanish. Being the best VCR repairman in the world is not a career path I would suggest.

Jobs We Won’t Want Robots to Do

There are jobs that, for a variety of reasons, we wouldn’t want a machine to do. This case is pretty straightforward. NFL football player, ballerina, spirit guide, priest, and actor, just to name a few. Additionally, there are jobs that incorporate some amount of nostalgia or quaintness, such as blacksmith or candlemaker.

Unpredictable Jobs

Some jobs are so unpredictable that you can’t write a manual on how to do, because the nature of job has inherent unpredictability. I have served as the CEO of several companies, and my job description was basically: Come in every morning and fix whatever broke and seize whatever opportunities presented themselves. Frankly, much of the time I just winged it. I remember one day I reviewed a lease agreement, brainstormed names for a new product, and captured a large rat that fell through a ceiling tile onto an employee’s desk. If there was a robot that could do all of that, I’d put down a deposit on it today.

Jobs That Need a High Social IQ

Some jobs that require high-level interaction with other people, and they usually need superior communication abilities as well. Event planner, public relations specialist, politician, hostage negotiator, and director of social media are just a few examples. Think of jobs that require empathy or outrage or passion.

Jobs Done On-Site

On-site jobs will be difficult to be done with robots. Robots work well in perfectly controlled environments, such as factories and warehouses, and not in ad hoc environments like your aunt Sue’s attic. Forest rangers and electricians are a couple of jobs like this that come to mind, but there are many more.

Jobs That Require Creativity or Abstract Thinking

It will be hard if not impossible for computers to be able to do jobs that require creativity or abstract thinking, because we don’t really even understand how humans do these things. Possible jobs include author (yay!), logo designer, composer, copywriter, brand strategist, and management consultant.

Jobs No One Has Thought of Yet

There are going to be innumerable new jobs created by all this new technology. Given that a huge number of current jobs didn’t exist before 2000, it stands to reason that many more new professions are just around the corner. The market research company Forrester forecasts that within the next decade, an astonishing 12.7 million new US jobs will be created building robots and the software that powers them.

Will there be some people in the future who are completely unable to compete with machines for work? It depends on which of the three possibilities comes to pass.

Possibility one is the easiest case. In that scenario, there is virtually nothing that the machines can’t do better than us. There may be a few jobs that for nostalgic reasons humans will still do, but for the rest of us, the period of humans having economic value will be over. It was Karl Marx who said that “the production of too many useful things results in too many useless people.” This would prove him correct in the extreme.

Possibility two involves unemployable people as well. The central idea behind this is that there will be a substantial number of low-skill workers who will not be able to compete with machines. In this view, the hierarchy of economic value in the future goes from skilled humans on top, then robots, then low-skilled workers.

In this scenario, there are a great many low-skill jobs that robots will soon be able to do. For each job replaced, the number of unemployed unskilled workers will increase and the number of unskilled jobs will decrease. So the world will have ever more unskilled workers competing for ever fewer unskilled jobs.

There are those who believe that we are already seeing this happen. To support their position, they point to the labor participation rate, which is the percentage of adults currently working. It peaked at 67 percent in 2000 and is down about four points now. The theory is that more people are completely withdrawing from the job market. Giving it all up, calling it quits. (The unemployment rate reflects only people who are looking for work, so these folks are not counted.)

It turns out that cyclical business cycle factors and the retirement of the baby boomers explain most of the drop, but not about 1 percent of it. Is that 1 percent a harbinger of things to come? Of course, we would expect the workforce participation number to decline as we get wealthier, right? Maybe both spouses no longer have to work, or perhaps someone’s bonus was so big this year that he or she is taking a year’s sabbatical. I think we would be hard-pressed to project onto this data the narrative that unemployable workers have lost their jobs and given up hope of getting new ones. It is notoriously hard to coax psychological conclusions from economic data.

Others point to the fact that since 2000, we have had increasing productivity coupled with flat wages and slow job growth. They see this combination as a sure sign that employers were growing their businesses by investing in technology and not people, lowering the demand for human labor. There are two problems with associating that economic data with robots taking jobs. First, the flat job market began abruptly in 2000 and lasted for about fifteen years, during which we had a couple of recessions, a financial crisis, growth in trade, and much more. It is not evident that automation was the underlying cause. Second, in the United States, 2015 had the largest growth in median income ever recorded and saw over three million people move out of poverty.

Finally, if possibility three happens, there will be no unemployable humans. But is this really possible? As mechanization and automation increase, surely there are some people who are left behind. Eventually some people can’t compete for work, right?

No. Assuming that a person is not afflicted with a debilitating physical or mental condition, there are no low-skilled humans. The difference between a human with an IQ of 90 and one with an IQ of 130 seems quite stark if they are playing Jeopardy!, but in reality, in the grand scheme of things, the difference is trivial. This idea is the basis of the well-known Polanyi paradox. In 1966, Michael Polanyi argued that there is a vast realm of human knowledge that consists of learning and skills that lie below our conscious thoughts. Think, for instance, about all the steps involved in baking a cake. Getting the dishes out, melting the butter, cracking the eggs, mixing the batter, frosting the cake, and so on. Virtually any human can do all of this, without even thinking about it. But a human’s abilities lie not just in making a cake, but in the ten thousand other things we can all do, like spot when our spouse is in a bad mood or brush our teeth or ride a bicycle. We are vast storehouses of ability, all of us. But because one person can do those ten thousand things and one person can do those ten thousand things—and knows about estate planning—we say one is low skilled and one is high skilled. But this is not the case, as they both have 99 percent or more of the same skills.

The idea that there are humans who cannot learn new jobs sells human potential short. The idea that the person doing a certain unskilled job is working at the limits of his or her ability is simply not true. Most people, in my experience, feel that they can do more than the job they have calls them to do. Given a chance to take on more challenging work in exchange for more responsibility and money, most people tend to say yes. People want meaning and purpose and, of course, higher wages, if they can get them. The only reason we use a person to shingle a roof is not because all that person can do is to shingle a roof, but because we haven’t invented a machine to shingle that roof. So while that roofer may have it in himself to manage 20 workers and come up with an aggressive plan for growth, well, the roof needs shingling and no one has built a machine to do so.

Imagine a person who mows lawns for a living. Let’s call him Jerry. Jerry graduated high school, but has no more education than that. Further, let’s say that someone develops a self-driving lawnmower that sells for a low price, and Jerry suddenly sees the bottom drop out of the lawn-mowing profession. What could he do?

A thousand things, actually. Remember, under the view of possibility three, all Jerry has to do is find a way to add value. Then he has a job. Jerry might, for instance, learn on the Internet how to plant and maintain grape arbors. That isn’t a big stretch, is it? I am not saying Jerry becomes a horticulturalist. He just reads enough to learn about how to plant and grow grapes. He then goes door-to-door with his message about the joys of growing your own grapes. Heck, I’d buy.

Then, 20 years later, Grape Arbor Robotics comes out with a robot that can plant vastly better arbors than Jerry can. So what does he do? He reads up on landscaping in the Victorian era. Then he goes door-to-door offering to plant historically accurate shrubs and flowers in historically accurate arrangements. Someday a robot will be invented to do that, but Jerry will have retired by then.

Who in the world could say Jerry is “unemployable”? He is powered by the most complex and versatile object in the known universe.

*Adapted from The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity by Byron Reese, published by Atria Books, an imprint of Simon & Schuster, Inc., Copyright © 2018 by Bryon Reese. Reprinted with permission.

RELATED ARTICLES

Bryon Reese is the CEO and publisher of GigaOm, an industry-leading technology research company. In addition, he is an award-winning author and speaker, as well as a futurist with a strong conviction that technology will help bring about a new golden age of humanity. Byron holds a number of technology patents and hosts two podcasts about artificial intelligence. He gives talks around the world about how technology is changing work, education, and culture. He has spoken about technology all over the world to combined audiences well in excess of 100,000.