What happens when robots and AI start taking all the jobs? Counter-intuitively, it’s the poor who will benefit the most. And it’s the upper-middle class who will complain the loudest.
Do robots eat poor people?
No. But they will nibble on the upper-middle class. And the reason is simple: the affluent have a larger gap between what they earn today and what would happen if their job function became obsolete.
When you lose your job involuntarily, the hit isn’t your entire income. Instead it’s your “opportunity cost” — the gap between your current job function and your next-best job function.
This opportunity cost gap is much higher among the affluent than it is among the low-skill. Meaning that, relatively speaking, it’s the affluent that will get hit by the roboto revolution. Not the low-skilled.
Indeed, if history is a guide, the low-skill will be the biggest winners from the coming robot revolution.
The upper-middle class emits most of what we read – intellectuals of any note are solidly entrenched in this class, after all. Professors, specialists, journalists and intellectuals don’t want to sound tacky pleading their own case, so they use the poor and low-skilled as sacrificial lambs in the Luddite debate. This sets us up for a good few decades of special-interest pleading clothed as concern for the poor.
Why can we be so confident that the poor will benefit? Because we’ve seen all this before. In fact, we’ve seen much worse. From that cousin of the robot, a little thing called agricultural mechanization.
In 1790’s America, agriculture made up some 90% of all American employment. Marching like the Four Horsemen of unemployment, agricultural mechanization slaughtered those jobs by the millions. Year-in and year-out for nearly two centuries. Starting with the iron plow, then the reaper and thresher, agriculture went from requiring 300 labor-hours per hundred bushels of wheat in 1830 to 90 hours in 1850, a 70% reduction. The job-killing march was relentless: 50 hours per hundred bushels by 1890, 20 hours by 1930, 5 hours by 1965 and, finally, 2 hours per hundred bushels by 1987.
Like a giant sausage machine, for two centuries 90% of all jobs were ground up and spit out in the form of shiny tractors and combines. Leaving the displaced and sadly obsolete workers to pick up the pieces of their shattered plowing careers. Of course, agricultural workers were typically the lowest-skilled in the entire economy. And they made up, from the start, some 90% of all American jobs. Agricultural mechanization, in its scope and in its ruthless industrial-scale destruction of low-skills jobs, makes robots and AI look like a tea party.
So what happened? Obviously, all the farm laborers starved and the American continent returned to nature.
Well, not exactly. Millions of displaced, obsolete, un-skilled farm laborers found new employment at much higher standards of living. The farms emptied as people flocked, first, to the factories and, later, to the emerging services economies in cities like New York, Los Angeles, San Francisco.
Their garages are stuffed to the gills with crap they don’t need, so obscenely rich are they by historical standards. They live in palatial homes with central heating, 24/7 electricity, indoor plumbing, free Skype calls worldwide, high-speed internet and, yes labor automation (dishwashers, washing machines). Luxuries that the richest in 1790’s America could only dream of. All while painfully aware that they really ought to go on a diet. Worrying about getting too fat would be something that history’s poor would find comically privileged.
A few years back in New York I overheard two guys unloading cinderblocks from a truck. One was talking about his recent vacation to Brazil. This would be a manual laborer, history’s hardest case, vacationing in Brazil.
This, then, is the poverty wrought by that job-crushing armageddon. And why did it turn out this way? Because of the mechanization itself. By making food so cheap, it made everything cheaper. Which we experienced in the form of rising quality of life, ranging from indoor plumbing to iPads and Brazil vacations.
So is the robot/AI revolution comparable to the mechanization of agriculture? Will it turn out as well as that great creative-destruction that brought cinderblock loaders vacationing in Brazil? Well, it’s actually much better than that for low-skill workers. For two reasons.
First, because robots and AI make a much wider range of stuff cheaper — not just food itself. And, second, because the jobs “targeted” by robots and AI are not just low-skill jobs.
And that’s where those intellectuals come in.
We want to back up and ask what’s a “job?” In a nutshell, a job happens when an entrepreneur realizes that resources can be combined and sold for more than they cost her. If some of those profitable resources happen to be labor then, voila, a job. The job-killing from robots or AI comes when robots are cheaper than that labor. In which case the entrepreneur buys the robot instead of hiring the worker.
So, what happens? Two things. First, the final product is cheaper. This must be so because the only reason the entrepreneur substituted robot for labor was because the robot was cheaper.
How much cheaper? We can only guess, since we have no robot takeover historical case studies. But taking agricultlure mechanization as a proxy gives us something like a 93% reduction in prices. For example, a 1795 British minimum wage worker earned enough in a day to buy 3.7 lbs of bread, enough to feed 2 people for a day. Meanwhile, a 2014 US minimum wage worker earns enough in a day to buy 32 pounds of bread, enough to feed a family of 4 for a week with a single days’ labor.
Let’s put this into concrete terms. Let’s say you’ve got a “good” industrial job today, paying $60k per year. You lose your job to a robot, and now you get a $30k waitress job instead. Or a $20k job as a babysitter. Well, if, as a result of all those robots, costs have declined by 93%, then that waitressing actually pays half a million dollars a year — about what a top Goldman Sachs trader makes. And that sitting job pays nearly $300,000 per year. Topping what a surgeon makes today.
Sound absurd? Absolutely. And a cinderblock loader vacationing in Brazil would sound absurd to anybody in 1880.
The key here, economically, is that a person who loses their job doesn’t just go on the scrap-bin. Their income doesn’t go to zero. Instead, it drops to next-best job. We can quantify this drop with actual figures: the average manufacturing hourly earnings in 2014 were $19.61 per hour. Let’s put in the worst-case scenario and imagine that worker’s skills are so low that their next-best opportunity is the rock-bottom lowest-paid jobs in the entire economy: personal care aide ($10.09/h), parking lot attendant ($10.26) or lifeguard ($10.05).
So, worst-case, a manufacturing worker drops to half their previous income, while mechanization suggests that real prices could fall by 15-fold. Leaving that displaced worker 7 times richer — 7-times being about the gap between Wal-Mart worker and distinguished Professor.
Now, what’s interesting for the coming Luddite debate is the distributional impact — the upper-middle classes are likely to be relative losers in the coming robot revolution. And the reason is very simple — the gap between their current job and their next-best job is larger than low-skill workers.
Consider a History Professor. He makes, say, $120k per year, or $80 per hour. When he’s displaced by tech, what does he do next? Well, he might not quite drop to home-care worker, but he might drop pretty close. Perhaps he’ll tutor children or start a blog. His income doesn’t drop by half, like the bumper-installer turned babysitter. It might drop by 3 or 5 times.
Yes, the cheaper goods will make our professor-turned-tutor better off. But he’s not making any more than that redneck bumper-installer anymore. In fact, she’s probably his neighbor now. This could be traumatic for our intellectual elite. And hence, I think, their concern. Not concern for the poor. Concern for their own spot in the pecking order.
Like the mechanization revolution, robots and AI will massively benefit today’s poor. And they’ll take the smart classes down a notch or two. Meaning we can look forward to a few decades of belly-aching from eloquent defenders of upper-middle class prestige.