It's a topic that has been played with by science fiction authors for quite some time now, and ever since the advent of artificial intelligence that came along with the first computers, the possibility that humans could one day be replaced by metal and silicon counterparts is becoming increasingly real.
Now, could humans ever be completely replaced by robots? It seems doubtful, at least anytime soon, but autonomous technology does keep imposing an ever-increasing threat to modern economic structures by replacing good ole hard working people with robots.
What Are Computers Good At?
Modern computers are pretty incredible displays of human ingenuity, and to most people their ability to "think" can only be described as "magic", but their manner of operation is not, indeed, magic.
When I took my first programming class in college, my professor described computers to us in this way: "they are the dumbest things you will ever interact with, but they are really fast."
And this is 100% true. Computers are made "smart" by smart humans who program them. They possess no intrinsic ability to "think", they can only be told to do very specific things. The power of computers comes in their ability to do things incredibly quickly.
Take the game of chess, for example, which has been used since the earliest days of AI as a test of computer abilities. It used to be that the best humans could consistently beat the best computers at chess, but now the tables have turned and the machines win every time.
Now, I don't know a whole lot about chess, but I did learn something about people who are really good at it: they don't think ahead (in the usual sense). Apparently novice to intermediate chess players approach the game by thinking about how their opponent can react to their move, and then how they can react to that move, etc. Many people assume that great chess players are simply thinking more moves ahead than an average person, but in reality they see the game completely differently. They don't think about discrete moves, they think about positions, and work on a level of abstraction above the average player. Any chess grandmasters, please let me know if I am way off on this.
If we wanted to, we can categorize chess players into two groups: the "elementary" players who just try to think a few moves ahead, and the "advanced" players who are able to abstract the game to a whole new level and see it completely differently.
If we wanted to program a computer to play chess, it would probably be ridiculously difficult to program an algorithm that approaches the game in the way that an "advanced" player would. Computers don't do well with abstract thinking, they are only good at, well, computing things. It would be far easier to program what is called an "n-ary" tree that goes through every possible move that could be made,and every possible counter-move, and so on, and choose the path that leads to success.
Now, while the second approach is easier, it also gets out of hand pretty quickly. There are an enormous number of possible moves in chess, and for each of those moves, an enormous number of counter-moves, etc. In the early days of computing, computers were neither very fast nor did they have very much memory, so algorithms such as the one described above would have to be truncated in order for the computer to decide on a move in the allotted amount of time. Nowadays, computers have gotten exponentially faster, and can do tremendously more calculations, making the above algorithm not only plausible, but effective.
This is just one example, but it is indicative of a greater theme of human and computing abilities. For humans to do things, there is usually what I call the "dumb-slow" or the "smart-fast" approach. The "dumb-slow" approach is anything that is conceptually easy to understand, but also tedious. The "smart-fast" approach utilizes some clever method to get the job done in considerably less time.
To understand this, consider that you are given the task of moving a bunch of heavy rocks from one side of your yard to the other. The "dumb-slow" way to do this would just be to carry each rock across the yard. A "smart-fast" way would be to go grab a wheelbarrow (which utilizes one of humanity's oldest pieces of technology: the wheel) and cart all the rocks across the yard in a fraction of time and with much less effort.
Most computer programs utilize the "dumb-slow" method to do things, i.e. they iterate through sets of data and perform tasks or make decisions based on what they see in the data. The only difference is that modern computers can perform billions of instructions per second, so in reality one can put computers into a new category: "dumb-fast".
Which Jobs Are At The Highest Risk of Being Replaced By Artificial Intelligence?
In short, if your job utilizes a lot of repetition of similar tasks, then you might want to think about having a backup career plan.
Assembly line workers were the obvious first choice for automation, as a single robotic arm could replace several employees and turn out products faster. If you've ever watched the show "How It's Made", you'll see that most of the work done in American factories is by robotic arms and other neat innovations, with humans doing some of the more specialized jobs in between.
It seems that the service industry may be the next target for automation. Instead of ordering a cheeseburger from a person, you just enter what you want into a computer kiosk and swipe your card. I've used these devices in Europe and...they aren't perfect. As with anything electronic, sometimes there can be confusion, so the places in Europe still had to have a few people watching over everyone (same with self-checkouts at grocery stores).
But, if government regulations continue to force businesses to pay employees more than they want to (or give more and more benefits), then they have greater incentive to minimize their human workforce. Even if businesses aren't forced to pay people more, there is still an incentive to automate a workforce. Robots don't slack off or call in sick (although they do break sometimes), so, while the startup costs are large, the long-term savings are often significant.
Humans as "Technology Babysitters"
People continue to freak out about how the threat of automation could put a lot of people out of work. The thing is, though, this wouldn't happen all at once. Technology has been replacing humans for centuries, and we've always managed to find a way.
Any new technology needs something of a "break-in period", a time when it can be fielded in a closely-monitored state while some of the bugs are worked out.
Case-in-point: I heard that at my job they decided to get a very large automated vacuum cleaner, basically a giant Roomba, to clean the floors. One would think this automation would put a janitor, who previously vacuumed the floors, out of business. Instead, there still has to be a janitor that walks around with the giant Roomba to make sure it doesn't get stuck anywhere, and set it free if it does.
In this example, the janitor didn't really lose their job, it was just re-purposed. Perhaps one day the janitor would lose their job once all the bugs were worked out with the automated vacuum cleaner, but it would definitely take some time.
This sort of thing would probably happen when automating the transportation industry. Many people fear that the increased belief in the performance of autonomous vehicles will put truck drivers and cab drivers out of work, which would be a significant number of people. I'm skeptical that this is going to happen any time soon.
Have you heard of trains?
Trains are massive and can cause considerable damage if they derail or crash into anything, but they also travel on fixed tracks and only need to regulate speed. The technology has existed for decades to replace train engineers with a computer system (we did a project on it in my undergraduate digital design course), but it still hasn't happened because of a combination of union deals, lack of financial incentive, and fears of a catastrophe. Some smaller trains today are completely automated, but all large freight trains still have people at the helm. From a business standpoint, you could probably save a few bucks by automating all freight trains, but if one had a crash, then those savings would be more than negated by the cost of lost cargo, lawsuits, and bad publicity.
The airline industry is more or less the same. Autopilots have existed in commercial airplanes for a few decades now, but for a while they were only able to regulate altitude and course, they couldn't take off or land a plane. Automatic landing systems have been developed, but are still not used all the time, and are closely monitored by pilots if they are used.
My guess is that the trucking industry would go the same direction. Even if we made autonomous vehicles a reality, they wouldn't replace all truck and taxi drivers overnight, there will probably be a long period of time where truck and taxi drivers simply babysit the systems and ensure they didn't go haywire.
Revenge of the Humans
If you're still scared that humans will be replaced by machines, consider this: there are several instances in which technology has existed that could replace humans with no safety drawbacks, but humans still have jobs in these fields.
One of these instances is at the grocery store. Self-checkouts have been a thing for years now at supermarkets, but most still have humans in the majority of their checkout lanes. Why?
Well, I generally don't use self-checkout because it's annoying. If I have produce, it takes forever for me to find the product codes, and even if I don't have produce, it's still just tedious to scan all the bar codes myself. To top it off, it seems like every time I try a self-checkout lane, it has some problem that the attendant needs to come over and help me with. It's generally around this time that I think to myself, "why didn't I just get someone to do all this for me in the first place?" While the job of grocery store clerk may seem easy, people who do this every day get really good at scanning products quickly, and most of the experienced ones have all the produce codes memorized, which makes the process much smoother.
This falls into a greater category of tasks known as "shadow work", or the work that everyday Americans have picked up as a result of greater automation. As it turns out, when you replace people with computers, you leave the consumer to do some of the work that people used to do for him. Craig Lambert's book Shadow Work: The Unpaid, Unseen Jobs That Fill Your Day points out how all these seemingly little tasks that we have picked up for ourselves are making us tired, depressed, and unmotivated.
It's true that businesses will constantly try to automate their workforce, but consumers may just decide that they prefer human interaction. Many companies are ditching automated customer service hotlines in favor of actual humans, because it has been shown to be profitable to do so.
While technology is here to stay, I'm optimistic that human beings will find a way to keep busy in spite of the growing threat of automation. Computers are incredible machines, but they may never be as good as humans when it comes to critical thinking and creativity.
So, if you think your job is going to be replaced by robots, it's never too late to start learning how to program said robots. As long as things run on software, we will need people to write software.
Which reminds me of a nerd joke:
"A doctor, civil engineer, and computer scientist were arguing over who had the oldest profession. The doctor said, 'God removed a rib from Adam, which was the first ever surgery, so doctor is the oldest profession!' The civil engineer said, 'But God created order out of chaos, the biggest civil engineering feat of all time, so civil engineer is the oldest profession!' The computer scientist chuckled at them both and said 'who do you think created the chaos?'"