Will machines proceed to take most of our jobs away from us over the next few decades? Whenever I suggest this I tend to get two very specific counter-replies. Here’s the first:
Reply #1: If we automate burger flipping, we’ll still need people to fix the burger-flipping machines.
Reply to the reply: There’s still a large net loss of jobs. If a burger flipping machine replaces 5 human burger flippers, and a service tech can provide ongoing service for 20 machines, then there are 100 burger-flipping jobs lost for every job re-gained in the form of burger-flipping-machine-service-technicians.
Of course, we should also account for those who will manufacture the machines, sell and market the machines, and perform administrative tasks for the company that makes the machines. But the jobs lost will probably still outweigh the jobs gained by a healthy factor.
And that leads to the second counter-reply:
Reply #2: When new technology displaces workers in one area of the economy, the economy always adjusts and finds places for the displaced workers in other sectors (or new sectors) of the economy.
When farm seeding machines, harvesters, fertilizers, and pesticides came along, it reduced the agricultural sector of the economy from about 50% of workers down to about 2%. Where did all those workers go? Well, they found other jobs. They became factory workers, clerical workers, scientists, and eventually burger flippers. It might not have happened instantly, but the economy eventually created new jobs to replace the old ones.
So we’re good, right? The economy always creates new jobs to replace old jobs.
Well, not so fast.
Reply to the reply: It all depends on gamma. Gamma? Yes, gamma? What’s gamma? I just made it up. Well, what is it? I’m glad you asked.
gamma = alpha/beta.
Ok, well then, what are alpha and beta?
Alpha is the rate at which the economy creates new human jobs.
Beta is the rate at which new technologies displace human jobs.
So gamma is the ratio of jobs being created to jobs being destroyed. If gamma > 1, then we’re probably in good shape (depending on how fast the labor pool itself is growing). If gamma < 1, then we’re probably in trouble.
And there are two reasons to think that gamma might turn permanently long-term negative, meaning the demand for human labor will more or less continue to fall steadily into the future (with some occasional reversals here and there).
- Our machines are getting closer and closer to being able to do anything we can do.
- The pace of innovation is increasing. (As computers have been getting more powerful and better connected, the number of things they are able to do has been growing exponentially.)
Those two facts put together suggest that we will eventually reach a point where a new human job can be replaced very soon after it is created. (Imagine that 6 months after a burger flipping machine is created, a burger-flipping-machine-repair-machine is created).
So, yes, the economy will continue to replace old jobs with new ones. But how many of those new jobs will be done by humans? Eventually, say we, almost none of them.
Do these two reasons hold water? That’s the question. Some would say that we are already seeing the effects of technology driving down the value of human labor. Wages have been stagnant for 40 years while productivity has doubled. And labor participation rates have fallen steeply over the last 10 years.
How much of that is due to automation (as opposed to, say, outsourcing, or reducing demand through trickle-down economics)? Who knows?
I don’t mean to settle the debate about machines taking our jobs here. I mean only to frame the debate by causing people to think about gamma, and to do away with two of the most common objections to the view that the machines are coming for our jobs.