Reading into headway made in the fields of Artificial Intelligence/Machine Learning, Robotics and understanding its impact on the future workforce, both near and distant, makes me question what will happen to the thousands, if not millions, of workers who will be displaced by intelligent systems that can perform monotonous jobs with great reliability and make precise decisions using learning algorithms applied to real-time data?
During the Industrial Revolution, 'Luddites' feared such a displacement from their work caused by newly developed machines used in cotton mills. Such resistance to new technology has been commonplace and so has been ridiculing of the so called 'Luddites' by people who embrace technology. The strongest argument against the Luddites is that technology eventually creates a new wave of job. For example, machines in cotton mills enabled quicker production thereby increasing the production capacity of the fabric. To make more profit from the expensive idle machine, the cotton mill produced more fabric than was required, resulting in cheaper selling price for fabric. The cheaper price of fabric would have caused an increase in demand. This additional demand had to be filled by new machines that had to be operated and hence the new wave of job. We will call this additional demand from price reduction facilitated by the technology as "new-wave-of-job". Hence, it is possible to argue that with training, displaced workers can eventually find new jobs created by the technology. In many cases, it is true that the quality of working condition improves, as technology takes care of the intensive mechanical part of a job leaving humans with the task of operating/managing the technology.
I happen to think the fear of Luddites is justified under the following conditions: (i) high reduction (>20%) in workforce requirement due to technology coupled with no "new-wave-of-job" (ii) self-learning algorithms with connectivity to real time data that eliminates the need for human problem solving and decision making. While the former factor displaces manual laborers out of the workforce, the latter displaces the white-collar workers who make everyday business decisions and create non-intelligent systems that need constant updates.
We see evidence for factor (i) mentioned above in many places where robots/autonomous systems are being deployed at a large scale. A current example is in the operation of Retail Warehouses where customer orders are picked and packed. Use of robots to manage these Warehouse operations are estimated to reduce workforce requirement by around 20%, with Amazon - a world leader in retail industry, leading the way. The use of robots help retailers such as Amazon with reducing the order processing time and thereby meet the shipping expectation of the customer. Here the technology is not going to reduce price of products, but only increase the pace at which orders arrive at existing customers. Hence, there is no "new-wave-of-job". To make matters worse in future, Warehouses will employ an intelligent system that tells workers the order in which they have to pick products so that the time and effort in processing a customer order is optimized. This amounts to worsening of the quality of work at Warehouses, as the volition of the worker is to be replaced with orders from a system on every task.
A future example with a more devastating impact on workforce is self-driving cars and trucks that are being keenly pursued by the auto industry. Self-driving vehicles are estimated to displace 5 million workers from active work force in the US, within the next decade. Once again, the self-driving vehicles are not going to result in "new-wave-of-job", but improve the pace and reliability of transporting people and goods.
Evidence for factor (ii) can be easily seen in the numerous efforts undertaken by machine learning scientists. The basic idea is to create an algorithm that will learn how to solve a problem. One such famous application is an algorithm that classifies the spam mails in our inbox. By reading the content of an email, the algorithm is able to determine with >99% accuracy if the email belongs to the spam folder. Another application that has been released in the recent past is the Google Photos that allows you to search your gallery of pictures for different shapes. For example, you can search you Gallery for pictures with noses and it will filter only those photos that have an nose like structure. One more example is the algorithm that determines the lane in which a self-drive vehicle has to be steered from the image collected by the camera in the vehicle. The beauty of these algorithms is that they do not need to be altered over time to do their jobs. Currently, efforts are underway to create an algorithm that would design a car based on the requirement of a customer. The pilot designs made by these algorithms seem far superior to car design created by professional designers. A current application of such a design created by an algorithm is the compartment divider between First Class and Economy in the newly released Boeing planes. In all the examples mentioned above, there is no need for human decision making, Also, since the algorithm has ability to learn, when there are small changes in the requirements, retraining the existing algorithm to meet the new requirement does not need a whole lot of human intelligence.
Coming back to the title of the blog:
Where will the next big wave of jobs come from?
Bill Gates, in one of his recent interviews, indicated that one new wave of job might come from providing personalized care/therapy to patients. Of course, none of us like to be nurses, which is one of the reasons for perennial shortage of nurses in the U.S.
If our grandparents had to deal with 1x change in technology during their lifetime,
our parents are currently having to deal with 4x change in technology during their lifetime,
we will have to deal with 16x change in technology during our lifetime,
The rate of application of technology is increasing in an exponential manner. Sometimes, I wonder what is it that I will end up doing at the end of my career, given these fast-paced changes in work environment and how it will impact my career path.
To say this is a problem of huge significance that needs more attention from all of us, is an understatement. I do not know enough Economics to understand the implications of such a big problem on our society. As a community, we need to think and talk about this issue more often.
P.S. Check out the intelligent robot developed in the lab of MIT that can learn to do simple tasks from co-workers. In the future, a laid-off worker will be training a new robot before leaving the job! That is what an entire season of 60 minute episode needs to focus on!!
