The Machine Learning Diet
When does our over-reliance on algorithms become the mental equivalent of too many potato chips?
By Xische Editorial, September 1, 2019
Source: Master Andrii/Shutterstock
What happens when machine learning becomes really good? That is the basic question unpacked in granular detail by Havard professor of internet law Jonathan Zittrain in a recent article for the New Yorker. It is a simple question that opens a pandora’s box of ethical dilemmas and questions for society. Chief among them is the issue of intellectual debt.
How many of us instinctively turn to Google for any question that pops into our heads? This is a form of intellectual debt in the information age. We have grown accustomed to outsourcing problem solving to the power of the internet. Instead of finding our way through a new city the old-fashioned way, we simply pull out our smartphones. Adding a tip to a restaurant bill is no longer a simple athematic challenge but another service for which the phone is summoned.
As we have written in the past, life is lost when we rely too heavily on machines for guidance. Getting lost in a new city raises the possibility of stumbling upon a hidden bookshop or beautiful cafe. But that is a thing of the past when our smartphones calculate the fastest way to our destinations and remove any mystery from the equation.
When it comes to the heavy-lifting of knowledge creation, we stand to lose even more. When you can find exactly what you are looking for you miss out on the unknown. When you visit a library (the ones with physical books), your eyes drift along with titles that might jog something deep in the brain. Amazon’s Kindle store has yet to recreate such an experience.
For Zittrain, the problem is particularly acute when it comes to machine learning. He writes:
“Intellectual debt has been confined to a few areas amenable to trial-and-error discovery, such as medicine. But that may be changing, as new techniques in artificial intelligence—specifically, machine learning—increase our collective intellectual credit line. Machine-learning systems work by identifying patterns in oceans of data. Using those patterns, they hazard answers to fuzzy, open-ended questions.
Provide a neural network with labelled pictures of cats and other, non-feline objects, and it will learn to distinguish cats from everything else; give it access to medical records, and it can attempt to predict a new hospital patient’s likelihood of dying. And yet, most machine-learning systems don’t uncover causal mechanisms. They are statistical-correlation engines.
They can’t explain why they think some patients are more likely to die, because they don’t “think” in any colloquial sense of the word—they only answer. As we begin to integrate their insights into our lives, we will, collectively, begin to rack up more and more intellectual debt.”
The last point of this long passage is critical. As machine learning systems get better and we incorporate them into our lives, we are at risk of losing our own thought processes. If this sounds alarming, think back to how many times you use Google to answer a question you could have solved on your own. Zittrain’s point is not that far-fetched.
This is not to say that machine learning is dangerous in and of itself. Quite the opposite. As a tool, machine learning can revolutionise society and human learning for the better. The medical field, for one, benefits dramatically from the insights of machine learning. The laborious work of government can be streamlined and perfected with the help of artificial intelligence, as clearly seen in the UAE example. These systems free up time and have the possibility to make life much happier for all members of society.
In order to realise the full benefits of the machine learning age, we need to inject mental fitness programmes into the process. That’s right, we need a machine learning diet. Just as one bag of potato chips is not going to harm you but a diet consisting of only potato chips could be extremely harmful, we need to approach machine learning from a “dieting” perspective. This starts with awareness.
We have talked about the danger of too much screen time for adults and children. Building on that, we need more discussion about the role of serendipity in our lives and the charms of using “old” technology once in a while. Not every question needs to be answered by Google. Looking for a word definition? Pick up a dictionary. Seeking inspiration? Pick up a book. Even the benefits of handwriting can come in handy here.
As long as we ensure that our traditional skills from writing to problem-solving are in sound working order, we can rest assured that our intellectual debt doesn’t get too out of control. Just as we are careful not to go into actual debt by carelessly using credit cards, we can ensure that we don’t regularly turn to machine learning systems to replace our thinking. This will prove harder and harder as machine learning gets better and better but we understand the contours of the problem. Now we just have to keep steady on our diets.
