A friend of mine and I are collaborating on a new book entitled Your Future Job: Building A Career In The New Normal. It's intended for college freshmen and their parents, though we hope it will help any intelligent young person. Maybe it's a bit too heavy on the economics for that audience, but we hope not.
We expect it to be around 100 pages long, and hope to have it completed by the end of Summer.
Here is a draft excerpt from Chapter 3 entitled What's New About Normal?
Of course computers can’t do things the same way that humans can. For example, it is now possible for a computer to drive a truck across the country without human assistance (as long as it stays on or near the Interstate; automated trucks can’t quite navigate city traffic yet). Humans do that because they have good vision--we can read signs that say “Kalamazoo--Next Right.” Computers can’t do that--computer vision is still very primitive compared to what humans can see. And for the longest time people thought that computers could never drive trucks.
But two things changed. First came GPS. That means the computer knows exactly where it is. And second came comprehensively detailed maps, such as supplied by Google. That means the computer knows what exit to take without having to read any signs. Indeed, every bend in the road is carefully mapped. The computer can navigate without seeing anything.
The truck-driving computer only needs to see the lane markers, and it needs to slow down or stop if something in front of it does the same. And here the automated truck has an advantage. It can use cameras on all sides of the truck--there are no blind spots. Further, it can use Lidar technology to determine how fast the car in front is moving--this is already in widespread use in the form of adaptive cruise control. The computer never falls asleep, is always fully attentive, and measures reaction time in milliseconds. It can drive 24 hours per day. The result is that automated trucks are safer, cheaper and faster than the human-driven sort.
It is only a matter of time before automated trucks dominate the nation’s highways. The consumer surplus will be huge--costs for things like fresh produce from California will decline dramatically. But millions of jobs will be lost. Over-the-road truck drivers will go the way of Pony Express riders.
And not just truck drivers. Computers, which are primitive today compared to what they’ll be ten or twenty years from now, will displace millions of workers in lots of professions. Brynjolfsson and McAfee cite the following table as a guide:
Table 1: Automation by job type
Not easily automated
Being automated now
Not easily automated
Manual routine jobs include a factory assembly-line worker. These jobs are already mostly gone, and the few that remain are disappearing. Similar jobs in fast-food places--flipping burgers--are also on the verge of automation. All of that can be done by machine.
Cognitive routine jobs are being automated right now. Dan’s father was a travel agent--a job that mostly doesn’t exist anymore. Accountants are increasingly displaced by software, such as Turbotax. Routine legal work is being automated--we’ll have more to say about that in a minute. Indeed, there is a long list of jobs that are in whole or in part on their way out: secretary, receptionist, purchasing agent, retail clerk, telemarketer, etc. Whole swaths of the pink collar and white collar workforce are on the bubble. Tom Joad knows exactly how they feel.
The non-routine manual jobs are a different story. Because computers continue to have relatively poor eyesight and hearing, and also are not particularly good at walking around, there are many things they can’t do. (Though there is much progress--see here and here.) Home health care aide jobs look to be pretty safe. Even if a computer had good vision, they still couldn't easily talk to patients.
Then there are the non-routine cognitive jobs, a category that traditionally includes professor and doctor. Tyler Cowen refers to them as jobs for the “cognitive elite,” by which he means people who can work with computers rather than against them. Certainly these are the jobs that many readers of this book aspire to. Table 1 suggests that these jobs will not be automated.
We don’t entirely agree with that conclusion, and that’s why we've put that table entry in italics. The category is more complicated than the simple description non-routine cognitive implies.
So before we tackle that last box in Table 1, let’s think about what computers will and will not be able to do. Inevitably, this implies peering into the future. We are very much aware of the fact that ten years ago people thought that computers could never drive a car. But in some ways that’s true--computers are nowhere near being able to drive a car the way people do. They simply can’t see well enough.
And here’s our prediction: computers will never possess vision, hearing, feel, smell, or taste that in any way rivals that of a human--at least not in your lifetime.
Take vision for example. Human color perception depends as much on psychology as on physics. Human vision does not just depend on the eye, but also on hormones. It’s a horrendously complicated process that evolved over a billion years (or, if you prefer, was created by God). A computer’s digital simulation of human vision will never duplicate the incredibly messy analog reality. Computers are valuable tools that can substantially augment human vision, but they are not even close to replacing us.
We’ll change our mind when a robot can play major league baseball, or compete in a PGA golf tournament.
Computers can flip burgers, but they will never be able to taste food and judge whether or not customers will enjoy it. The skilled chef will always have a job. A computer can diagnose cancer, but only a skilled human being can gather the necessary data from a patient, either by observation or conversation. A lot of specialist physicians will be unemployed while the primary care doctor is still working.
Computers can replace the cashier and ring up purchases, but they will never rival the good salesman. If you already know what kind of car you want, you can order it online. But if you’re not sure, then the personable car salesman is going to play an important part in your decision. Likewise with the real estate agent, or even those travel agents that still exist.
In the last few years a new word has entered the language: big data. This refers to the analysis of a large amount of data, from which conclusions can be drawn. The Jeopardy-playing computer, Watson, and millions and millions of trivia data stored in its memory, along with an algorithm to retrieve them on command. Its command of big data allowed it to beat a human Jeopardy player. Computers just memorize all the answers. That’s the way they drive a truck--they don’t need to see the road because they’ve memorized it. Computers have a better memory than you do.
Computers can also do math better than you can. Math really means any kind of logical manipulation--not just with numbers. They can manipulate words just as easily, and they can do it very quickly. While chess-playing computers have memorized all possible moves for the first 15 or 20 moves into the game, after that they use logic to calculate further moves. As computers get bigger they’ll simply memorize whole games and dispense with the math. Then it will be impossible to take them by surprise.
Here’s where the “cognitive elite” are going to run into trouble. Any job that requires a large amount of memorization will be computerized. As mentioned, that includes many doctors. They spend years in medical school memorizing diseases, symptoms and treatments. This is not time well-spent. Computers will take over that part of the physician’s job. Similarly, professors pride themselves as fonts of knowledge. Sorry Prof, but the Internet knows more than you do.
So here is what it means for you:
- Your future career depends on skills much more than knowledge. Computers know things, but they can’t do things.
- The more specialized you are, the more your job is at risk for automation. The general practitioner will be employed long after the oncology specialist is laid off.
- Jobs that employ the five senses in a sophisticated way are more secure than those that don’t. The chef and the artist have secure jobs. The laboratory technician maybe not so much.
- People skills are really important.