Automation and income inequality
Tech Is Splitting the U.S. Work Force in Two
Still beating this drum. People need to understand this dynamic because Trump and his followers either don't have that understanding, or are deflecting, when they blame it all on immigrants. John Oliver recently did a piece on automation -- also worth watching.
Let me try and explain what I'm doing in my current job. I work for Amazon (standard disclaimer: I don't speak for them). Amazon used to make all its money from selling to customers, but now makes billions from AWS, Amazon Web Services. This is a tech company leasing its tech to other companies. What for? So those companies can do their own buying and selling online, and whatever else they do. It's safe to assume that part of what they're doing is automating their own processes.
Netflix is a well-known AWS customer. If you're on Netflix, go find Hasan Minhaj's excellent show Patriot Act. There's an episode about Amazon, in which--to turn the whole thing on its head--he also explains AWS.
Given its success, it's no surprise that AWS is expanding all the time. It is forever adding new services to the menu, and it's also expanding physically, spreading to new regions of the world. Those services, as you can understand, are computers talking to computers over the Internet. Teams of software developers write those services and maintain them, but the services themselves are programs, i.e. they are automated. So are the services that let you buy access to services. That means a company can expand its use of AWS to meet rising demand, in a way that is also automated. Still with me?
All those teams of AWS developers roll out copies of their services to the new regions that come online. This requires going through a number of steps: acquire some computers from the cloud, deploy software to those computers, set up new databases, and so on. (All while sitting at their desks, usually thousands of miles off.) Then a new region comes online, and then another. You don't want your team of developers repeating the same steps over and over: it's a waste of time, and they'll inevitably make mistakes. So what do you do? You automate again. That is what the team I'm on is doing. We are writing programs to help teams deploy copies of their programs that service corporate customers' programs that service human customers. This is automation cubed, or maybe hypercubed; I've already lost count.
When this article asks, "Could automation go too far?" -- I think that's a pointless question. John Oliver made the right call: Automation is not going to stop, not so long as it makes economic sense. If a computer can do a job, it can almost certainly do it more cheaply than a hired human worker. Version 1.0 might be worse at it, but version 2.0 will be along shortly.
John Oliver's story went into some of the nuance, as he so masterfully does. Millions of truck driver jobs? Soon to be replaced by AI drivers. Can you put all those former truck drivers to work as software developers? Some, certainly, but not all will be equally skilled at it. I like to think that anybody with an interest can learn to write a computer program, but we must assume that not everybody is equally comfortable with mental abstractions. (Non-programmers often conflate software development with math. They're both abstract, so you could consider the two fields to be neighbors, but I do very little actual math day-to-day.)
As a side point, I disagree with how this article draws a firm line between the "highly educated" and "less educated". I don't regularly advertise the fact, but I do not have a college degree. I have a rather mentally challenging software development job, I have twelve years under my belt at Amazon, I have a six figure salary, I have a house in Hawai'i that my wife and I bought last year, and what I do not have is a college degree. Most of what I learned about how to program computers, prior to getting my first job doing it, was self-taught. I owe my parents an unpayable debt of gratitude for buying me computers as a kid. If you have the ability to think abstractly and enjoy it, and can afford a computer and Internet access, I believe you can teach yourself to program. Most tech shops are desperate for more genuine talent, spending many hours separating those who can from those who can't, and believe me, many of those who can't do have a college degree. That means the piece of paper is not all-important. Many of those companies, the smarter ones, will give you a chance to prove yourself without it. I hope that fact helps a few people who can't afford to risk taking on mountains of student loan debt (another societal disaster which John Oliver has talked about).
But back to the main point: Automation is indeed eliminating jobs, and though it's creating new ones, we can't assume everyone is equally capable of doing them. Those of us on the upside of the trend must be willing to help those on the downside. A lot of us are. Much of the push to consider Universal Basic Income (UBI) is flowing from people in the tech sector. We think something like that, funded by higher taxes on the rich and the well-off, is the solution... not a big ineffective wall.
"The forecast of an America where robots do all the work while humans live off some yet-to-be-invented welfare program may be a Silicon Valley pipe dream," says this article, rather dismissively I felt. If it's a pipe dream, then I say we'd better get started on laying that pipe... er, no, but you know what I mean. It's only a dream until we decide to do it. I believe we need to do it.
In the near future, automation might endanger as many as half the jobs -- so says the report John Oliver referenced on his show. He went on to explain why it's not that simple. Here's another reason which he didn't go into: Over a longer time span, that 50% figure is probably way too low. Unless you believe there's something inherently magical or otherworldly about the human brain (and I do not), someday, learning computers in robot bodies will be able to do anything and everything humans can do. On the day they can do my job, when the computers can program themselves and the robots can build and repair themselves, that's when we hit 100%, the end of the line. By then we must have transitioned to an all-new economy based on something other than human labor, that ensures everyone equal access to the boundless fruits of all that robot labor. And a voice in how we use it.
Comments
Post a Comment