Overautomation – the Value of Returning to Manual Work

I regularly post about the advantages of using natural (as opposed to artificial) intelligence in the workplace.  I also carefully say that there are two kinds of work: routine work that should be automated, and unpredictable work that should not be automated, and it should be fairly easy to distinguish the two.  But is it?  Toyota is taking some surprising actions putting intelligent workers back in positions formerly thought to be routine.  It turns out those positions weren’t as routine as they originally thought.

Automation normally replaces intelligent workers with machinery.  Let’s face it, for much of history most people have held jobs that required little or no thinking.  Automating routine work is a good thing, because it frees people to participate in the more challenging knowledge work.

I often use the automobile factory floor as an example of a place to automate as much as possible.  The factory is a controlled environment, where external unpredictable influences can be mostly eliminated.  Inside that environment, one should automate as much as possible.

In the surprising article “Toyota is becoming more efficient by replacing robots with humans” and from Bloomberg “‘Gods’ Make Comeback at Toyota as Humans Steal Jobs From Robots” we learn that Toyota is actually reversing this trend.  That is, they are putting people back in to routine jobs that could be done by robots. Or could they?

The reason for doing this is very sound:  only by actually doing the job can you understand the job, and suggest improvements to the job.  It is not good enough to simply watch the robots and try to find areas of waste.  It suggests that there is more to the job than what can be seen.  That tacit knowledge comes from the manual work, as you focus your action toward an explicit goal, the mind is always working on other, tacit objectives.  For these unconscious processes to work, you need time, and you need engagement.

The article cites some dramatic improvements.  This does not by any means indicate the end of automation.  Suggestions from these workers will be used to improve the automation elsewhere.  The automation remains great for doing the exact same thing over and over.  The full job is not merely doing the same thing over and over.  The world changes, and the factory can not be completely isolated from it, and it needs to change as well.  The machines that automate the factory lines do not suggest ways to improve that assembly line.  Instead, you need some number of intelligent people working the same job.

Perhaps this will start a new trend and a new job role.  For every business process analyst, there will be a “master worker” — someone who actually does the entire job manually. The master worker not only understands the process in detail, but actually manually performs it on a small number of cases. Instead of calling a web service, the master worker actually carries the work from place to place. Sound crazy?  I don’t know.  Maybe.

However, one thing we should not be surprised about:  it is not always easy to distinguish routine work from knowledge work.  Sometimes there is a knowledge aspect of a job that is not obvious.  Reductionists tend to assume the job is simpler than it really is and will tend to automate some things that should not be.  So as automation proceeds, it will be natural to experience “overautomation” when you go too far.  We should be looking for overautomation, and then de-automating some things.  It is natural that that through this process of automating, and de-automating, we will explore the boundary of what can and can not be automated, and eventually find the optimal amount of automation.