Automation and Exponential Growth: Not a Problem. Until it is.


Four years ago I wrote about the concern I had about automation and the problems it would cause to society.  It seemed as logical then as it does now that at some point computers and machines will be able to do every task better and more efficiently than humans.  Since then, I have noted that many prominent academics, like Stephen Hawking, have come to the same conclusions.

Yet despite being in good company, I have had a very difficult time convincing others of this obvious and serious obstacle.  I often run into some variation of rebuttal in which I am simply making the luddite fallacy.  The idea that technological change will result in long-term unemployment has been made for over 200 years.  And like the hundreds of prophets who have predicted that ‘the end of days is coming,’ anyone who has made the prediction the ‘end of work is coming’ has been wrong. While technological change might result in some short-term unemployment in that industry, the long-term effects to society have always been beneficial to society.

Given the history of technological change and the skepticism by many of my colleagues, I have often seriously considered that I have been wrong.  But using the past to predict the future is not always an effective strategy.  In a related example, an individual concerned about the jobs of horses[i] may have been concerned about the invention of the ‘horseless carriage’ (car) and the tractor.  Individuals trying to assuage those concerns could have pointed to the fact that technology in the past had only increased the number of job horses had.  The horse population boomed with the advent of every new technology; that is, until they didn’t.

But why have people had such a hard time with the argument that technological advancement is a growing concern?  I recently came to an answer that I think helps to explain it.   Consider the following simple problem:

A lily-pad on a pond doubles in size every day.  After 100 days the lily-pad will completely cover the pond.  On what day will the lily-pad cover 50% of the lake?

Did you answer it?

Was your answer 50 days?

Well if so, you are wrong.  The correct answer is day ninety-nine.  Don’t feel bad, the above problem deals with exponential growth, something that humans have a hard time intuitively grasping.  The question above is relatively easy but produces an intuitively wrong answer.  A related problem is used on the cognitive reflection test, a very simple task that has good predictive power in explaining a variety of cognitive abilities.

Now imagine you told someone that the lily-pad was going to cover the pond after 100 days.  On day 90 it would only be about .1% of the pond.  Trying to explain to someone that has been watching it grow every day, that in 10 days’ time, this little lily-pad that has barely grown at all, will cover 100% of the lake would be laughable. People who were warning others about this problem since day 1, would be told just how wrong they were, that is, until they weren’t.

This same intuition may be responsible for why so many have difficulty accepting the growing concern of automation and technological unemployment.  Although human potential and technological improvements are very difficult to properly quantify, we know that processing speed  has been increasing at an exponential rate[ii] and serves as a good proxy for technological advancement.  We may not know just how far away we are from ‘Day 100’ (e.g. a time when computers and machines can do every job better than humans) but we know we will get there. And just like the lily-pad, technological speed and subsequent technological unemployment will grow exponentially.  With each doubling the associated problems also double. Past unemployment problems associated with automation and technology may have seemed small because they were but they will only continue to grow, exponentially.

We need to recognize that with exponential growth, problems don’t seem like problems, until they are.


[i] Of course, humans aren’t horses. We are smarter and can learn to do a variety of complicated tasks.  But that only means that it just takes a more advanced technology to replace human employment. In addition, although horses may have seen a huge drop in their ‘employment’, the overall quality of life for the typical horse has dramatically improved.  Horses are often fed, sheltered and given healthcare.  We don’t simply stop caring for horses because they are no longer useful.  This is similar to the premise of universal basic income.

[ii] One possible point of contention is that that Moore’s Law (exponential growth of processing speed) is only an observation not a rule.  There may be a point in which we start slowing down in technological advancement. On the other hand, we may start speeding up as well. Regardless, as long as technological advancement doesn’t come to a stop, we will still reach a point where  technology can beat humans.

Pictures taken from: from



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s