Join us for a live webinar, PISA in Perspective: Cultivating Mathematical Mindsets, March 18 at 1pm ET. Register here.

Cross-posted at Education Week

I’m really looking forward to our sailing trip next month. I’ve been wondering lately what you have been thinking about what you want to do after high school.  At 16, this must seem to you to be a long way away, but it is going to sneak up on you quickly.  I know you plan to go to college, but what sort of college?  How would you choose?  Is there something you think you might want to do after college that might influence that choice?  Why that choice and not some other choice?  It is, I think, a good thing we’ll be out on the North Atlantic for a week.  These are not easy or simple questions.  By the way, I may be able to help you think about your answers, but they will have to be your answers, not mine.

As I thought about this trip and these questions, I began to wonder how I, after having lived a rather full life, would answer them now.  Which is not to say that you should answer them the same way, but my answers to the questions I have asked you might at least be food for thought.

The fact is that when I went to Brown all those years ago, I gave no thought at all to what I wanted to do when I graduated.  Relatively few people went to college, and, because the economy was growing rapidly and college grads were in great demand, I didn’t think I had to worry about it.  As it turned out, I was right, but you are in a very different position.  The situation is reversed.  There are lots more college graduates and the job market favors employers, not the applicants for jobs.

This should affect both the field you choose to study and the college you choose to attend.  I wrote a blog a few weeks ago that made the case for you to get the strongest liberal arts education you can get and your father and I can afford.  I won’t repeat it here, but the gist is simple enough.  Many of the good jobs out there today won’t be there in 10 or 15 years, so, if you don’t have what it takes to learn fast and adapt quickly you will be cooked, no matter what you choose to do.  Many colleges that advertise themselves as bastions of liberal education are not worth the powder to blow them to Hades, but some offer the best strategy you can find anywhere to build the knowledge, skill and deep understanding you will need to support a lifetime of learning.

Your uncle did exactly that. Your uncle had decided to be an electrical engineer, but the college within the university he attended required him to complete a demanding liberal arts curriculum based on the old Columbia University Western Civ program before he started his engineering studies.  He could have transferred to another college at his university and saved the time and money that the additional year would cost.  But he didn’t.  It was one of the best decisions he ever made.

But what then?  You have been a very good math student for years now.  That is a skill you can build on, if you want to.  There are lots of ways to do that.  Here are some things I would take into consideration.

There are lots of jobs for mathematicians that pay well.  They are in great demand, as are people trained in other occupations that require strong mathematical skills.  That is partly because many fields that did not require much mathematical skill have now become highly quantitative in their orientation.  Mathematics transformed the field of economics over the last fifty years, but that is only one of countless fields that have gone through such transformations.

But be careful.  Many people—not just front-line workers, but professionals with advanced degrees, too—will be put out of business by intelligent devices in the years after you graduate.  You don’t want to be one of them.  People with strong mathematics skills are strong candidates for instant obsolescence.  After all, mathematical calculation is a very strong suit of computers and computer-like devices.

One defense against the possibility of being made instantly redundant by intelligent machines is to be on the other side of the fence—that is, to be the person designing the new machines and applications, not the person whose job is taken by the newly designed machines.  Even there, you will have to be wary.

I’ll give you an example.  One of the most interesting of the emerging advanced technologies for automation is “deep learning.”  When educators use that term, they mean deep learning by people.  When computer scientists use it, they are talking about the machines, not the people, doing the deep learning.

Learning in the brain is a matter of making connections, zillions of them.  Some years ago, computer scientists invented something they called neural networks to mimic the networks of neural connections in the brain.  They have gotten very good at this, as they have learned how to put one layer of simulated neurons on top of another in one computing environment.


They have now married this thinking capacity to advanced possibilities for gathering and organizing data and for visual pattern recognition.  Consider now what happens when your dad takes you to the doctor and the doctor tries to figure out what is wrong with you.  The doctor is trying to remember what she learned in medical school, what she read more recently in the journals, what she was told the other day by the drug firm rep who stopped by and so on.  She might have seen only 30 or so people who presented symptoms similar to the ones you describe.  It could easily be that your case, though similar on the surface, is not actually related to the causes of the problem in those cases.  Some other question that your doctor failed to ask might have shown that, and enabled her to accurately diagnose your problem, but your doctor did not know that.

But now, using these new forms of neural networks, coupling them to vast databases of medical histories and photos, as well as the greatly enhanced analytical capacity of these machines, the modern computer scientist can literally teach the machine how to learn to recognize and accurately diagnose a whole range of diseases far more accurately than most doctors.  Nurses located in remote rural villages will have more and better information and analytical capacity available to them than most doctors do today.

We’ll still need doctors, but maybe not so many.  Maybe, you say, the job you want is the job of the people designing the new neural networks and the software that connects them to the databases and analytical capacity that will enable them to make sense of the data banks to which they will have access.

Good idea.  But here is something for you to think about.  It turns out that the engineers have now designed neural networks that are learning to design neural networks.  Hmmmmmm…..

Maybe you don’t want to be the engineer who designs the neural networks after all.  You might want to be the engineer who designs the neural network that designs the neural networks.  But, hey, you are 16 years old.  Who knows how these machines will evolve by the time you graduate with a degree in engineering?


There are now only about half a dozen laboratories in the world that are now working on advanced applications of neural networks.  If it were me, that’s where I would want to be working: in labs that are working at the frontier in a field that has a future, and there are quite a few of those.  But, if you choose to do that, you will have to bear a few things in mind.  You will have to be very, very good at what you do.  You will have to be prepared to work very hard.  You will have to be not only technically proficient, but also someone who your colleagues like to work with.  And you will have to learn constantly, without stopping, for the rest of your life.

But maybe you would prefer to be a professional sailor, sailing wealthy people’s boats to the locations they would like to sail in.  You won’t make much money and the retirement benefits are nonexistent, but the life is great while it lasts. There is, after all, nothing, nothing at all, as Kenneth Grahame said, like messing about in boats.