by Jay Johansen | Feb 28, 2016
People who fancy themselves to have insight into trends and technology often say that computers and robots will soon be able to do anything that people do and do it better. Some predict this will happen within just five or ten years. A few optimists say this will lead to a utopia where people live in ease and comfort while machines do all the work. They are greatly outnumbered by pessimists who say this will lead to massive unemployment and poverty.
Will either scenario really happen?
People have been predicting this since the dawn of the Industrial Revolution. In 1811 a movement started in Britain, known as the "Luddites", who broke into factories and destroyed machines because they were afraid that these machines were taking away jobs. Yet today, 250 years later, most people still have jobs. The massive unemployment that the Luddites feared never happened.
There are two basic reasons generally given of why machines will take away jobs.
One: Machines increase productivity, and increasing productivity reduces the need for workers.
Today unemployment is about 5%. So 95 people can produce all the food and clothes and houses and so forth required to support 100 people. But as technology advances, machines do more and more of the work. Say that soon machines can do half the work. Then it will take only 50 people -- and a bunch of machines -- to produce all the food and clothes and houses and so forth required to support 100 people. The other 50 will then be unemployed.
The flaw to this reasoning is that it assumes that human desires are constant. If, adjusting for inflation, your grandfather lived on $20,000 per year, and you make $50,000 per year, than it inevitably follows that you should be saving $30,000 per year, or giving it to charity, or something. But in fact, that isn't what happens. As incomes go up, we consume more. When I was a boy air conditioning was a luxury for the rich. Today 87% of American homes have AC. When I was a boy most families had one car. Today my family of 3 has 3 cars and I don't think we're that unusual. Cell phones weren't even invented then, no one had them. (A small number of rich people or people who really needed them had radio phones.) Today it seems everyone over 12 has one. Etc. I'm sure you could extend this list yourself.
In a thriving economy, when new machines make a job obsolete, people don't become permanently unemployed. They get new jobs producing products that were rare or unknown before. And the standard of living for everyone goes up a little.
Two: When machines can do the job better than people, businesses will use machines rather than people.
As the technology of robots and computers improves, they become more and more able to do jobs that used to require people. The Industrial Revolution replaced human muscle power with machines. The Computer Revolution is replacing human brain power with machines. Soon there will be no jobs left that humans can do better than machines. Or at least there will be only a few jobs requiring very special skills and training, that most people will not be qualified to do.
But again, you can see the flaw if you compare what is happening today to what happened in the past. Yes, factories today are now filled with machines rather than manual laborers. But there are still plenty of human beings in factories, running the machines and doing jobs that the machines can't do. Yes, today computers are doing a lot of things that used to require human intelligence, like keeping track of bank balances and taking orders. But there are still plenty of people working in those fields, running the computers and doing the things the computers can't do.
When ATMs were invented, there were many predictions that bank tellers would all lose their jobs. In fact, the first ATMs came out about 1970. According to the Bureau of Labor Statistics, in 1970 there were 290,000 people employed as bank tellers in the United States. In 2010, there were 580,000.
But, advocates of the AI Doom theory say, today is different. The Industrial Revolution replaced human muscle power with machines. So people got jobs that required intelligence instead of muscle. (Not necessarily genius-level intelligence, but the ability to read and write and do arithmetic and make decisions.) But the coming AI Revolution will replace human brain power with computers. This time, there will be nowhere left for people to go. It's not the same. In the past people could always retreat to jobs that machines still couldn't do. But when machines can do anything that a person can do, there will be no place left to retreat.
But any such scenario is much, much farther away that the doom-criers would have you believe. Machines are wonderful at mindless, repetitive processes. But unless and until an Artificial Intelligence is invented that truly rivals human intelligence, machines simply cannot do jobs that require creativity or flexibility. And such an AI is nowhere in sight. Many people who don't know a lot about computers are very impressed with Artificial Intelligence. But whatever you can say about the current technology of AI, one thing you can say is: They do not operate very much like human beings. They are not "creative" in any sense.
I once made a computer game of running for president. (Lest you wonder: it sold several hundred copies. It did not make me a millionaire.) Like in many computer games, the human player competed against an AI. How did it work? It consisted of a mass of detailed rules and formulas. For example, in real life, a politician's campaign manager might say, "Let's concentrate our efforts on states where we are not so far ahead that we'll probably win even if we do nothing, nor so far behind that we'll likely lose no matter what, but on places where it's close. Oh, and we want to concentrate on the states with the most electoral votes." So I put such a strategy into the AI in my game. But it had to be reduced to a formula: For each state, calculate how far ahead or behind in the polls. Take this number minus 50%, square it, and subtract the result from 100. Multiply the result by the number of votes. Then find the state where the result of this formula is largest. Of course that was one rule of many. There were dozens of such basic formulas throughout the AI and thousands of details. My point is: such AIs do not invent strategies. They carry out strategies invented by people. Computers can do the raw arithmetic faster and more accurately than any human. But they can't decide what arithmetic needs to be done.
I read a book on AI once where the author discussed "expert systems", AIs that are programmed to identify automobile mechanical problems or diagnose illnesses and so on. And, the writer claimed, this problem had largely been solved. Computer systems have been developed that can reproduce the thinking processes of experts. There's just one problem remaining, he said: The human experts who must be consulted to formulate the rules for the AI are often unable to express their knowledge in a way that can be programmed into the computer. I laughed. His position is that computers can think just like humans ... except that no one knows how to describe human thinking in a way that fits into his computer program. He started out saying that he has solved the problem of how to make computers think like people, and ends up complaining that there's the small catch that people don't think enough like computers to make it work.
It may be that someday people will invent a true Artificial Intelligence, that is, a machine that can really think just like a human being, or like a human being but better. But if so, such machines will bear little resemblance to 21st century computers.
I saw a video on this subject recently where they talked about all the jobs that they thought could be done by computers and robots with technology that will be available within the next few years. For example, they said, self-driving cars are in serious development now. Very soon we will not need truck drivers or taxi drivers. They then pointed to statistics on how many people are employed in such jobs and said, See, that's over 45% of the population. Within a few years, 45% of the population will be unemployed. I'd say they were being over-optimistic about many of the technologies they discussed, but let's grant their assessment for the sake of argument.
In 1840, 69% of Americans were farmers. Today about 1% of Americans are farmers. That's why unemployment today is over 68%. But it's not, is it? Why not? Because this analysis is simple-minded. Just because someone's current job is made obsolete by technology doesn't mean that he is doomed to unemployment for the rest of his life. It is even more fallacious to suppose that if one person loses his job today because of advancing technology, that therefore there must be one person in every generation for the rest of history who is unemployed. It's like they think that everyone is born with some specific job pre-programmed into him, and if that job is not available, he will be unemployed. If in some alternate universe I would have been a mastodon herder, that because the mastodons are extinct in our world, therefore I am doomed to a lifetime of unemployment and poverty.
As a side note, I wonder if this is a telling point about the thinking of people who warn about this scenario. They say that machines will replace people, because they think that people are just like machines. A machine is made to perform one specific task. Like a carburetor is made to mix gasoline with air and feed it to the engine's cylinders. But then fuel injection was invented so carburetors became obsolete. Therefore all the carburetors were thrown away and there is little place for carburetors in the world any more. Likewise, a certain person was born to be a truck driver. Soon self-driving cars will make truck drivers obsolete. Therefore all the truck drivers, and all future people who are born to be truck drivers, will be unemployed, useless, and thrown away. The flaw, of course, is that no one is "born to be a truck driver". People are flexible and adaptable and can perform many functions and learn to perform many more. Unlike machines.
The same video ridiculed the idea that people can always move on to truly creative tasks that computers can't do. Yes, they conceded, maybe AIs today can't be truly creative, maybe there are some things that still can only be done by people, like writing poetry. But, they said, "we can't have a poetry economy".
I'd say, "Umm, yes we can." I doubt we'll ever reach a stage where literally the only thing that people can do that computers can't is the truly creative. But even if we do, why couldn't we have an economy where people did only creative work and left all the mundane work to machines? "Creativity" in the sense we're using it here doesn't just mean writing poetry. It means anything that requires coming up with a new idea, or putting ideas together in new ways. It includes writing poetry and novels, inventing new machines, and making scientific discoveries. But it also includes inventing a new board game, developing a marketing plan, devising a coherent political platform, or thinking of new things that computers can be programmed to do. A computer might be a valuable tool in doing any of these things. But it isn't going to think of the new ideas itself: it can only manipulate ideas programmed into it by people.
Perhaps I should add: Yes, people do sometimes lose their jobs because of advancing technology. I'm not denying that. What I'm denying is that this is a condition that will inevitably last for the rest of the person's life and for all future generations.
The process is certainly not always painless. Some people who lose their jobs in such circumstances quickly find another and it's just a brief inconvenience. More often, it's a serious problem for a few months or a year before he finds a new job and gets back on his feet. But someone who has spent a lifetime acquiring and developing a skill that is no longer relevant may be forced to start over again in a new career, likely at lower pay, and with a loss of prestige and self-confidence. It's easy for the person who is not in this situation to say, "Hey, that's tough, but you'll find another job. And you know in the long run it's good for the economy." Most people probably get new jobs fast enough that in the long run, they benefit from the advancing technology just like the general population. But there are certainly individuals who do not. For them, to say, "Yes, you lost your job and will never have another job that is as good. But that's okay because your neighbors are better off for it" -- probably not much consolation.
Photo: © Scherbet | Dreamstime.com - Robotic Arm Showing Victory Photo
© 2016 by Jay Johansen