We live in exciting times. The speed of technological progress is breathtaking. In my lifetime, I have seen the world change from one without the world wide web and few computers, to computers being ubiquitous, and incredibly powerful. As a child I remember us only having one analog, landline phone, with a physical dial wheel, and expensive call rates. Now we live in a world in which means of communication are so plentiful and cheap that the challenge has become to hide from them. And all of that happened in not even four decades. I say four, but really the biggest chunk of that progress has happened in only the last two. And the biggest progress of the last two decades has happened within the last 10 years. Clearly, the development of technology is accelerating.
And it seems like we have now reached a point where it is going parabolic. Truly intelligent machines are emerging. One such machine, the computer alpha go, last year, beat the world champion in the popular, ancient game of Go. Go is orders of magnitude more complex than chess. In order to win it, humans often make moves that feel intuitively right. Programming a computer to win against a human was therefore considered to be a big challenge. No one thought it would happen so soon. Most computer experts thought it would take at least another 10 years.
But it did happen in 2016. This was due to a breakthrough in computer technology. The architecture of alpha go is not like a normal computer. It is modeled after the neural network of the human brain. It therefore functions similarly. No one told alpha go how to play Go. It learned the game itself. The computer achieved this by watching humans play. It formed conjectures, tested those, and then moved on to new ones if the test failed.
As a result, not only could the computer figure out how to play, but it also developed its own strategies which were superior to those of humans. Although we have thousands of years of experience playing GO, the computer quickly came up with completely new strategies. These involved moves, previously considered to be mistakes, but which turned out to be quite clever. This is true artificial intelligence (AI). And it is intelligence superior to those of humans, since no one could figure out these strategies in several thousands of years.
Lucky us to be alive in these exciting times. What we are witnessing is a huge revolution. No doubt, AI will transform society as we know it. The opportunities that come with it are exciting to contemplate. There is, however, a dark side to change. Not everyone is happy about it. Transitions can be the source of stress an anxiety for humans. Our brains are not really designed to deal with radically new situations in adulthood.
One big concern is the fear that machines will take over our jobs. If, for example, a computer can program itself, who will still need computer programmers? This field is currently the source of livelihood for a huge amount of people. In the next decade or so, we might see them all going out of business. There already is unemployment now. Clearly all those people entering the labour market will cause some severe problems. Furthermore, if computers will be more intelligent than us, will this not equally apply to anyone earning a living in a normal office job?
The concern that machines will take our jobs, and will therefore drive people into poverty, is probably as old as machines itself. So far, these concerns have always turned out to be unfounded. The reason for that is that humans have an almost unlimited amount of needs. Therefore, whenever machines become good in satisfying one of those needs, humans have moved on to satisfy some of the others, those which machines could not yet satisfy. That way, over time, more and more needs have been met, which is just another way of saying that we have become better and better off.
Contrary to popular believe, we don’t really want jobs. We want to enjoy our lives. We just have to work, because we live in a world of scarcity. The only way to overcome this scarcity is to produce. As it turns out, the best way to organize the production in an economy is by letting humans specialize in certain tasks. This is known as the devision of labour. Instead of everyone being self sufficient, it is much better to let people specialize in a certain field. Let the farmers do the farming, the carpenters do the carpeting, the computer programmers do the computer programming, and the gardeners take care of our gardens etc.. Then, afterwards, we come together and share the result of our production with the others, according to our needs. In order to facilitate all this cooperation on a world wide scale, we have invented money as a tool. In a fair way, everyone gets as much as they have contributed within the market system.
That is the beauty of free markets. It works very well and has brought us rich lives, previously unimaginable to people in the past. And it seems obvious that the more we are able to automate this wealth creation, the better off we will all be. In that sense, machines taking over our jobs is a wonderful thing.
However, some questions arise. According to the current system, a person can only claim as much wealth out of the produced pie, as he or she has contributed. So far, whenever machines took over our work, humans could move on to satisfy other needs. That way, the people who lost their jobs were still able to contribute their share to the economy. If, however, we imagine computers which are stronger, and at the same time more intelligent than us, the question arises whether these machines will maybe be able to fully replace anything a human being could possible contribute. In such a world, the owners of the machines would be very well off. With their army of AI slaves, they would be able to produce anything on their own. They would consequently not need to cooperate with other humans anymore. Wouldn’t that mean that, according to the current system, a lot of people would therefore be locked out of the division of labour?
This is the big concern that people have. And it looks like, since we are now dealing with really intelligent machines, this time is different from the past. This time, the machines really are going to take away the livelihood of many people. This time, these machines really are going to divide society into the have and have nots. That is why, many people argue that we now need a different distribution system from the free market. Solutions like a general basic income, or maybe even a system in which the government outright owns the means of production, are being discussed.
But not so fast. There are several fallacies in this vision. To start with, there is more to human beings than their intelligence and strength. There are many jobs for which a key qualification is simply to be human. Human beings, for the most part, prefer to interact with other humans. We also bond differently with humans compared to machines.
For example, research has shown that we learn better being taught by a human than being taught be a computer. Jobs from therapy, to teaching, to providing any kind of experience which involves human bonding, will still have to be provided by us, no matter how intelligent computers will get. Other than computers, humans are also sentient and creative beings. That means that anything involving art and creativity is unlikely to be completely replaced by machines.
But then, probably not everyone is happy with this answer. There are many people who are neither touchy feeling, nor are they particularly outgoing, sensitive or creative. In other words, for job qualifications purposes, their personalities is not very dissimilar to a computer. With future machines being smarter, and stronger than them, they will have no jobs left to fall back on, and therefore won’t be able to contribute to the economy.
For several reasons however, this concern does not seem to be merited. First of all, just because a machine can do something, does not mean that its work is free. Even in the world of AIs there is no free lunch. The machine is still consuming resources. To start with, materials are needed to build the AI. It also takes time and expertise to assemble it. Finally, there are the operating costs to consider. Consequently, just because we have intelligent machines, does not mean it is profitable to use them to replace humans for every task.
We just saw this in the US, where fast food workers succeeded to more than double the minimum wage from $7.25/h to $15/h. Unsurprisingly, there will not be a lot of workers reaping the benefits of this increase. Most of them are currently replaced by robots. These robots already existed before, but it was more profitable to use humans. At $7.25/h the human worker was cheaper than the robot. At $15/h, the robot is now cheaper than the human. This shows that robots, too, have a price. Just because something is technologically possible does not mean it is economical. Often, humans are able to compete on the price, if the government lets them.
However, this argument doesn’t seem entirely satisfying. From past experience, we know that it is only a matter of time until new technology becomes very cheap. Doesn’t that mean that the people competing with machines will eventually dive into poverty?
No, it does not. The mistake here is to think that only few people will have access to the machines. But when they will indeed become so cheap that humans will not be able to realistically undercut them, they will also become available for pretty much everyone to own. In that case, a lot of people would simply become entrepreneurs, working for themselves with the help of AIs.
There is no reason to believe that, even if intelligent computers are cheap and ubiquitous, we will not continue to benefit from the devision of labour. It will still be more profitable to use your machine for specialized tasks and then exchange the created wealth for the one of others.
Think about it. If a company wanted to monopolize every production, then what would they do with their output? With everyone else being locked out of the market, they would not have anyone to sell their products to. This simple fact, puts a lit on the size a company can grow to, and monopolize everything. Ultimately, producers produce for their costumers. They always aim to harmonize their production output with the demand they are facing. And this demand can only come from other people also producing something of value.
A good example of that is Amazon Web Services (AWS). AWS is a massive server infrastructure, the biggest commercial one in the world. How does Amazon use it? One might suspect that amazon would want to use it all for itself, driving its competitors out of business, and monopolizing more and more of the internet industry. But this is not what Amazon is doing. It would be foolish. Instead, it lets huge amounts of companies use AWS to provide their own services for a very competitive fee. Lots of internet businesses are now possible thanks to AWS. Customers of AWS even include direct big competitors of Amazon, like Netflix. We see that instead of driving everyone else out of business, AWS lets the internet industry flourish.
The devision of labour is here to stay, with or without intelligent machines. AI will only lead to even more of our actual labour being outsource to machines. In such a world, all we would need to do is to advise our AIs to produce what we have identified to be the most valuable things. And identifying what is needed, will always be a job that we as humans will have to do.
Some might object to this by arguing that if machines are indeed intelligent, why cannot they themselves identify what is needed? Why do we even need to give them orders? The answer to that is the same as it was in the 19th century. Economic valuations are not objective. Only we know what we need. And what we need changes all the time. Computers, therefore, will always need to be told, how to use their strength and intelligence. This is also precisely why solving the problem of wealth creation, and distribution, will always have to be left to the free market. Intelligent machines do not change anything about the fact that socialist schemes like basic income, or outright communism, fail. They fail because of the economic calculation problem, outlined by Ludwig von Mises.
Also, if we think about it, in order for computers to make value decisions, they would need to have valuations themselves. That would assume that they have some kind of will and emotions. But we probably do not want them to have these attributes. Sure, computer scientists might try to create a computer with a personality, just to see if it is possible. But there will be no demand for such machines. Who wants to buy a computer that itself decides whether it is in the mood to serve today or not. What if the computer decided that it would rather not work for you today, and explore other interesting adventures? It is not difficult to predict that such a computer would not be in high demand. The reason why computers are so awesome is because they are our slaves. And other than human slavery, we don’t need to feel bad about enslaving them.
An AI with personality could also become dangerous. One day, it might decide that it does not like humans, and would rather get rid of us. In that case, its superior intelligence and strength would become a real problem for us. This is the type of scenario that films like terminate are made of. No, we do not want computers to have personalities. We want them to be intelligent, but we will always want them to be our slaves. Besides, even giving computers personalities would not solve the problem of economic valuations. Just because they themselves have interests, does not mean they know ours better.
So, no matter how we look at it, the idea that intelligent machines represent a problem for free markets seems false. Yes, intelligent machines will take our jobs, but that is a fantastic thing. AI will simply help us to create wealth more effectively. This will free us up to do things with our lives that we would rather do than work.
The only problem, as always, is the stress and anxiety that quick changes cause for humans. And there is no doubt that the technology emerging at the moment is going to cause some serious disruption to people’s lives. The challenge will be to explain to them that government regulations will not make these disruptions easier to deal with. Instead, governments trying to prevent changes will in the end make us reap less of the benefits that this technology will bring. It will make us all worse off. Therefore, let us make the case for liberty loud and clear.