You Can Lose Your Job to Automation Even If Your Job Cannot Be Automated

Get Excited

Are you excited about automation yet? You should be. Here's a great video to get us started:

But I'm tired of all these misconceptions about automation replacing jobs! Consider these statements:

• Jobs are safe when no robot can do 100% of the job.
• Jobs are safe when they pay very little.
• Jobs are safe when they pay a lot.
• "Human touch" jobs are safe.
• Unemployment is bad.
• Jobs are safe when experts at those jobs think they cannot be done by a robot.
• Engineering and computer programming won't be automated soon.

Do you believe any of these? If so, I've written this post specially for you!

1. No robot can do 100% of my job

So what?

Consider this hospital robot. It rolls around and rides elevators to transport medication, food, and linens around hospitals. But this is just a small part of the duties of a medical orderly. People may say that a medical orderly will never lose their job to automation because robots cannot do 100% of their tasks.

Lets say just 5% of an orderly's time is spent on those tasks a robot can now do. Meaning 1 in 20 of their work time has been automated. That also means 95% of this human job cannot be done by a robot. Nobody is losing their job, right?

Wrong! If we have 20 human orderlies, now we only need 19 orderlies and a robot. The hospital can maintain the same level of service by letting someone go. Even though 95% of the job totally cannot be done by robots yet, a medical orderly can lose their job to automation right now.

The calculations here are simplistic but they should illustrate a point: you can lose your job to automation even if your job cannot be automated.

2. I earn too little for my job to be automated

The hardest jobs to automate are sloppy, random, and improvisational. There are low wage and high wage jobs for either category. Wage is disconnected from automation.

Consider this cool olive harvester.

The very low wages paid to unskilled human harvesters didn't stop anyone from creating that machine. Also, automated checkouts at stores aren't perfect but many people use them. So those low wage cashier jobs have also been lost to automation. Both these jobs are very repetitive and easy to automate.

Now consider a solo concierge and janitor for a school. This is an incredibly challenging job to automate - I say this as someone who has dabbled in machine learning recently and loves learning about it. Here are some tasks a janitor must do:

• mopping up a kid's puke
• duct taping a shattered window to make it safe until replacement
• shutting off a toilet's water after it gets possessed by the devil
• moving all chairs to another room
• cleaning a pool that a kid puked in

Even the coolest humanoid robots we have today are nowhere near this.

If all we know about a job is that it has a low wage, we can't say one way or another it if will be automated soon.

3. I earn too much for my job to be automated

Drivers of gigantic mining trucks can get paid over 100,000$ a year. They need special training and must live in the middle of nowhere. We're starting to see this high wage job disappear to self driving trucks. This job may pay a lot but driving a truck in a mining pit is very repetitive compared to a school janitor. We don't pay people higher wages because their job is hard to automate. We pay people more because of the desirability of the job and its skill requirements.

The only difference with high wages is that there's more financial incentive to automate it. So high paying jobs may be harder to automate (or appear harder) but more resources are also going into automating it.

High wage jobs are neither safe nor vulnerable to automation.

4. My job is safe because of the human touch

Are you a live performer, waiter, third wave barista, live musician, or analog artist? Do you think you offer something special, intimate, artistic, or human that no robot can ever do? Well then you probably really do offer something special.

You can still be automated, even if you cannot be replaced. Even if all people preferred a human waiter, many would still opt for automation if the robot waiter is cheaper or faster. Maybe there's something objectively better about live music - but many people choose to listen to recordings now instead of live music. Most bars would prefer a live band to perform all their music if it cost the same as a laptop with a playlist. But of course it doesn't. That preference for the human touch is usually not materializing into human jobs.

5. Unemployment is bad

Being wealthy lets people do what they like. You can enjoy any hobby or work on any project all day. Maybe you'd just choose to be on permanent vacation. In a world of full automation, then everyone can be wealthy in this way. We won't be wealthy compared to each other, but wealthy in our freedom to do whatever we like.

In this science fiction dreamland there would be nearly 100% unemployment. Not the economic collapse type of unemployment, but technological unemployment. This type of unemployment is good. Note how present day economies and media outlets are not tracking rates of "good" unemployment. Instead, all unemployment is seen as bad.

If we can reach the right balance between social programs and entrepreneurial competition, unemployment can be celebrated.

6. Human Experts

According to a Pew Research Center survey most Americans predict that within 50 years robots and computers will do much of the work we do today. But most Americans also think their jobs are safe. Hilarious!

I once read an interview with the union leader of mining truck drivers. He has twenty years experience as a driver. He says that driving in "hard" dirt pits has been automated, but that he drives in the more difficult "soft" dirt pits. Therefore, his job is safe.

What I find most remarkable is that he likely knows nothing about computer vision and machine learning. And yet he still claims to be an authority on the automation of his job - based on his experience as a human worker. Don't fall into that same trap.

Being an expert at your job does not mean you understand how to automate it.

7. Engineers and Computer Programmers

People often insist that computer programming and engineering (especially robotics engineering) will be some of the last jobs to be automated. But consider this picture taken in the 1980's at Boeing:

This picture was taken before engineering design software was created. It's pretty incredible and shows how much the career of design in engineering has changed. In the past, these engineers were experts in technical drawing and the use of specialized drawing tools. That job has now been automated by software.

Similarly, computer programming is being automated. Even though the number of computer programming jobs has increased massively in the past couple decades, we'd need an even larger workforce of computer programmers if we didn't share code. Open source projects and code sharing automates the job of computer programming. Whenever a computer programmer uses a library instead of writing it from scratch, they've used technology in a way that eliminates the need for another computer programming job.

Remember misconception 1: "no robot can do 100% of my job". As engineering software improves, and open source solutions expand, both these jobs will also be threatened by automation. This occurs even if parts of these jobs truly are the most difficult of all to automate.

What Jobs Can be Automated?

Try not to pay attention to trendy news headlines like "Will a robot take my job?" from the BBC, or "Will your job be done by a machine" from NPR, or "42% of Canadian jobs at high risk of being automated". They are jam packed with these misconceptions, especially 1 and 4. Instead, consider these questions:

• Can a robot or computer do part of my job?
• If a human had a robot body, could it do my job?
• If a computer had a human body, could it do my job? If so, what are the most flexible, sensitive, articulate, and precise robots today?
• Have similar jobs just been automated?

And always remember: You can lose your job to automation even if your job cannot be automated.