Back at the turn of the century, I had a job working in the IT department of a Big Bank. The section I worked in was Mortgage Exceptions. My group’s task was to route exceptions that could not be addressed by the mortgage processing system to a human being for resolution.
Most of the intelligence required to execute the Big Bank’s mortgage business processes was automated by the mortgage system. Each month the system processed millions of mortgage payments. The systems sent out the bills for payment. And, when a payment came in, a human operator entered the check payment into the system. (This was before the days when direct withdrawal was the norm.) Most people paid their bills. So, for the most part, the system chugged along without a hitch. But every so often there was a problem the system could not handle—an exception, if you will.
Examples of exceptions were accounts that were significantly past due, account numbers that were unknown to the system, customer complaints and bounced checks. There was a whole array of stuff that could—and did—go wrong that the software could not handle. Thus, the exception ended up in front of a human being. The human was the ultimate exception handler.
I worked on that system more than 15 years ago. I think it’s safe to say that today the bank’s system intelligence is a lot smarter. I imagine that today the IT group I worked in is more concerned with writing more robust, automated exception-handling than routing problems to humans.
That’s the way technology goes. Systems get smarter and, in many cases, become better than humans when it comes to doing work that is well-known and repetitive. The more the machine can do, the more the machine will do. Businesses like machines—always have, always will. Automation is a good investment.
Those of you who have been following my writing on DevOps.com know that I am very concerned with the impact of automation on human employment. I saw the writing on the wall when I read the Ball State study. As time goes on, more machines will be doing more work. Less human labor will be required. What will those displaced humans do?
Many believe that automation will create new jobs for humans. In some cases, it will. As I said earlier, I am sure the people who were writing the routing code at the Big Bank I worked at years ago are now writing more robust exception handlers that require less human interaction for resolution. Those few developers writing the exception handlers are probably doing just fine. The developers who were writing simple if-then rules, with no ambition otherwise, are probably gone. Also, I’ll bet the the Big Bank has embraced the DevOps sensibility, which probably has resulted in those with lesser talent in terms of implementing automation being shown the door. As for those humans who were entering checks into the system and turning delinquencies over to collection agencies, their number is probably now a handful, if any. Hopefully, the poor soul who was hired as a teller out of high school and worked her way into mortgage exception resolution adapted, despite the odds.
As I said previously, many believe that automation will create new jobs. There always will be new things to do or things that need to be done that machines can’t do. Fair enough. Still, for me it’s not about new jobs being created. It’s about the number of new jobs that will be created.
Many of the new jobs will require a lot more smarts than did the jobs they are replacing. And my gut tells me that, while there will be new jobs, there will be fewer than the number of old jobs eliminated. And these new jobs will fall into two categories: “Cool New Work” and “Doing Work the Machines Have Yet to Master.” The “Cool New Work” is just that. “Doing Work the Machines Have Yet to Master” is handling exceptions, the work I used to route to humans back at the Big Bank. Except this time, as systems become smarter, many of the exceptions these systems throw will become a lot more difficult for humans to resolve. The systems will throw fewer errors, but when they do, watch out!
Exception-handling has been in the realm of advanced human activity since the days when the ancient Mayans needed to figure out a way to build pyramids that didn’t fall down. A lot of buildings fell down before a few stood up. Eventually, they figured it out. Exceptions gave way to learning, and today we have the art and science of architecture. The industry has matured. Today, constructing a building is more about process and technology than figuring out what went wrong with the last building.
So, too, will it be with automation exceptions. Just as over time fewer pyramids collapsed, with automated systems fewer exceptions will be raised and fewer humans will be needed to figure out what went wrong. Then, we’ll be back to where we seem to be going: more machines doing more general work, a smaller number of very intelligent people doing more advanced work, and the rest of us doing the tasks the machines can’t do yet—aka, the exceptions. We’ll be working on exceptions until the time that the thinking machines will be able to resolve those exceptions. Then what? Dunno.
Now, don’t get me wrong. Technology is good; automation is good. There’s more to life than spending most of one’s waking hours entering check deposits into a mortgage system. Humans were meant for more. But, when earning a living means being really smart or being just one step ahead of a machine, it’s going to be a real challenge for the Average Joe to the find opportunities to earn a buck, let alone find a job. How will we meet this challenge? This I leave for you to determine.