A new study from Oxford says that 47% of the jobs in America are at risk of automation. There is a lot of fear that a job automated is equivalent to a job eliminated. It is the same fear that fueled the Luddites, however history shows that fear to be misplaced then, as it is now. Automation drives a transformation of the workplace, not an elimination.
The report by Carl Frey and Michael Osborne at Oxford University is called “The Future of Employment: How Susceptible are Jobs to Computerisation?” says:
… Based on these estimates, we examine expected
impacts of future computerisation on US labour market outcomes, with the primary objective of analysing the number of jobs at risk and the relationship between an occupation’s probability of computerisation, wages and educational attainment. According to our estimates, about 47 percent of total US employment is at risk. We further provide evidence that wages and educational attainment exhibit a strong negative relationship with an occupation’s probability of computerisation.
Pretty scary, right? Almost half the workforce — gone. In the early 19th century automated weaving looms were introduced in England, and in 1830 the followers of Ned Ludd protested (and rioted) on the theory that jobs had been taken away by automation. A large part of the workforce was doing the kinds of things that no longer needed to be manually performed. The argument was that without these jobs, the economy would collapse as these people sat around without any work, and without any money to purchase the goods from other industries. Martin Ford’s book “The Lights in the Tunnel” gives an updated, Silicon Valley view of the same fear of people being displaced from work. He talks about Luddites, but insists that ‘this time it is different.’
History actually shows a different story. While automation undoubtedly has a short term effect of displacing particular workers who have grown accustomed to their present habits, it does not show any trend of decreasing the overall size of the economy, or of decreasing the number of jobs available to the population. One only needs to consider that around 98% of the manual labor performed in ancient Egypt is now automated, but that does not mean that 98% of the workforce is now sitting idle. The automated looms opened up new opportunities to design fabric, while the increased production opened up new opportunities to distribute and retail the fabric and clothes.
The fallacy seems to come from thinking that the job categories that we have today are the only job categories that we will ever have. Advances in technology eliminate job categories all the time. The electronic printer has completely eliminated the need for a movable type printing press like the kind that Ben Franklin used to operate. That is right, nobody is wiping ink on their aprons as they stack little bits of metal with raised letters on the end into trays any more. Word processors have eliminated the need for typing pools in offices. Email has eliminated the people would would pick up memos from your out-box and deliver them to someone else’s inbox. Spreadsheets have eliminated the need for people to add things up with calculators, while calculators had in their day eliminated the need to do calculations by hand. The automobile eliminated the need for buggy-whip manufacturers.
Transformation of the Workplace
What does happen, is that when a routine job is automated, it opens the possibility for other non-routine jobs. When a robot is placed on an automobile assembly line, it displaces an assembly line worker, but it creates a need for a programmer as well as a new support chain for the machine. I am not saying it is one-for-one (i.e. there is one programmer for every assembly line worker replaced) but simply that new job categories are created that the Luddites could never imagine, and therefor did not enter their calculations about the economy. Desktop publishing displaced ink-stained typesetters, but opened the way for three-D rendered graphics and self-published books.
Automation does cause a change in the aggregate nature of work. Only routine work can be automated. The new jobs created tend to be less routine, and more knowledge oriented. This has the overall effect that the working population is tending to be more knowledge workers today than ever before. I have used these graphics to illustrate this trend:
This is intended to represent office workers at the dawn of the Internet age where are majority of office work is routine work, with a smaller number of people who are primarily engaged in knowledge work. Many of these routine jobs have been automated or otherwise eliminated, and we might represent the current economy more like this:
In spite of all the automation that has appeared in the office, and also in spite of the large unemployment numbers today, according to US labor statistics, the total number of human workers in 2013 is 20% more than in 1995, from 114 million to around 139 million. Automation over that period has not decreased the amount of employment.
However, because many of the routine jobs have been automated, we find that more people are doing knowledge work today. Estimates of the actual percentage vary, but it could be as high as 50% of the labor today is knowledge work.
Time for Knowledge Workers
Thus the automation of routine tasks by BPM and other automation techniques is not in general eliminating work, but instead transforming the work that people do from routine to knowledge work. You don’t have to manually print letters and put them in envelopes, but you still have to decide what to write in your letters. 3D printers automatically make intricate physical object, but you still have to decide what to make and design the actual shape. This transformation has positive and negative aspects. Knowledge work is generally more interesting and satisfying than routine work. However, it is also more challenging and stressful as well. Let’s face it: routine work is a lot easier than knowledge work.
This represents a huge opportunity for Adaptive Case Management as a technology to support innovative, creative work. As automation continues to displace routine work, possibly up to 47% of the work being done today, the demand for new technologies to meet the need of knowledge workers will dramatically increase.
The young Italian Federico Pistono wrote a book with the title Robots Will Steal Your Job, But That’s OK, argueing why now more and more old jobs will be destroyed than new jobs can be created.
Here is his speach on TED hold in Vienna this year:
When this reduction of jobs will happen, why should we try to work so much hours per day as today. Shouldn´t we use this time for helping people arround us?
Thanks for the reference. I might get time to read this. I spent a long time considering this subject when I reviewed Martin Ford’s book what took a similar position. The reason I believe that new jobs will be invented as fast as old ones are automated, is because there continues to be need for specialists. Cancer needs a cure. Mental illness needs study and understanding. There are so many unsolved problems. Imagine every job we know today perfectly automated. But you still have cancer, and so some people are surely going to pay others to dedicate their time to study the disease. The economy is nothing more than doing work which you have become specialized in for someone else, in exchange for money which is a substitute for work you need others to do. Until every possible human need is met, there will always be work. However, I would agree with Pistono that when every possible human need is met, we will no longer need to work, however, that will not happen in the next few centuries.
Keith, great article and the triangle graphics puts a visual perspective to the change in work that we’ve seen of the past few years. Can I suggest that you post the link that you did with BPM.com where you discussed the graph. I think it is worth listening to and also to see how this knowledge style work require a more adaptive approach.
Thanks for the suggestion. The talk was called “Strengthen Organizational Agility with the Latest Advances in Case Management” and the video of this can be found at:
(requires free registration)
One thing I miss in your analysis, is joblevel vs skilllevel. It is true that automation also creates jobs, but these are typically jobs for people with higher education and more specialized skills. For a period of time, lower educated people were still able to shift to new niches of work in industrial, administrative or service jobs, but I predict this to become more and more difficult. It will lead to an already emerging class of lowpaid flexworkers.
It might be well possible that we as society need to ponder on new questions as ‘what is the meaning of a life without work’. Will we create an upperclass of workers and underclass of the less workless humans, or can we come to other societies, embracing and fairly distributing wealth, free time and power?
Yes. The knowledge work is generally associated with higher skill levels, which often, but not necessarily, implies higher schooling level. At the time of the Luddite rebellion literacy rate was around 50%, and today it is around 100%. Clearly, you need to read to to just about any job today. This is a disruptive change and while the picture show the population elevated, you must understand that there are 20 years between the pictures, and many people from 1995 have “aged out” of the working population by 2015. Some people are replaced by automation and simply are unable to learn new skills. However, the main point is about how the population changes, with new workers tending to start in knowledge work positions in greater and greater numbers. This population change justifies a change in technology to support it.
I would urge everyone not to conflate the idea of education with knowledge work. A receptionist, for example, might be an excellent knowledge worker – not because they have a PhD in running the reception, but because they watch, listen, and learn who comes and who goes, and they learn and use the patterns which are not rote. Knowledge work has to do with how “predictable & repeatable” a job is. No matter how skilled you are, if the job is predictable it might some day be automated. For example some kinds of surgery will soon be automated, but for now takes a person with 20 years of education.
Distribution of wealth is another thing entirely, and I recommend the book “How Nations Fail” (Acemoglu and Robinson) for a comparison of political & economic systems which are inclusive vs. extractive. Imagine a world where everyone has a 3D printer in the garage and all production is decentralized. Your neighbor makes Lamborghini replicas – not because he went to school to learn this, but instead because he absolutely loves Lamborghinis, and consequently his work has become valuable. That is a wild speculative view, but I am not yet convinced that automation necessarily leads to extractive economies that cause a big divide. I believe there are enough “tough” problems, like how to solve health problems, including mental health problems, and stop crime, to keep the population fully employed when all the routine tasks are automated.
Pingback: Automation elevating workers, not eliminating _ thinking matters about information and communication technology