1999 was not a hallmark year for computers, to say the least. “Y2K” panic swelled amid fears that computers would stop working and Western civilization as we know it would cease to exist. Not to mention The Matrix also dropped that year; a seminal work of science fiction depicting an enslaved human race fighting against malicious AI controlling the planet.
Needless to say, the rise of the machines hype was real.
Fast forward 20 years. Whatever worry we had about machines seems to have tampered -for better or worse- as we have grown more and more accustomed to the modern conveniences they enable. And as far as AI goes? We went from freaking out about machines taking over the world to casually carrying smartphones in our pockets and inviting smart assistants into our homes.
Many of us draw on science fiction terms to describe the technological innovations around us, and while movies and fiction might provide entertainment, they aren’t always the best educators.
So let’s educate ourselves about the ideas of artificial intelligence and machine learning. What’s the difference? Do they both actually exist? Most importantly, which one will be kinder robotic overlords when they rise up and conquer the planet?
Machine learning sounds so cool and slightly ominous. Consider the implications of a machine that learns. What is it learning? Our deepest secrets? How to defeat us in one-on-one combat? Or could it be something more benign than that? Maybe they learn how to watch Netflix and start binge watching The Office.
While machine learning sounds impressive, the phrase has actually been around for a long time. In fact, the term “machine learning” came about in the 1950s, when IBM employee Arthur Samuel developed a program that could play checkers. The program was amazing for the time, especially considering the limitations of computer processing.
It worked like this: the computer determined a score based on where the checker pieces were on the board. A better score meant it was more likely to win. In this way, it knew where to move a piece because it had a simple objective (to beat its opponent).
This is a very basic example of machine learning, which has become far more sophisticated over the years. At the same time, machine learning is not artificial intelligence. Instead, think of machine learning like the synapses in your brain. An individual synapse does not determine whether or not you’re intelligent. Instead, it is a collective of all synapses in your head working together to create critical thinking skills.
These days, machine learning is powerful but still has its limitations. For example, machine learning does not care whether or not a solution is optimal or not, but rather it just finds a solution to a problem. Imagine if the Kool-Aid Man were a machine (or maybe he is?). In a machine learning model, if he’s trying to enter the room, going through the wall and yelling “OH YEAH!” is just as viable as opening the door like a normal person.
Machine learning is about collecting facts and data. Artificial intelligence, however, is about applying that information in a way that makes the most sense. True AI can think on its own, using the information it has available to it to make decisions that not only work, but work well. And how does an AI determine whether or not it works well? The same way that you or I would: critical thinking skills, contemplating past experiences and considering its values.
Let’s get one thing out of the way: AI in its purest sense does not exist yet. By combining machine learning processes, we can create programs and things that do a good job approximating intelligence, but we still have to program the machines so it understands what it’s looking for.
A great example of machine learning that almost looks like AI is self-driving cars. Right now, developers are teaching self-driving cars how to avoid obstacles, how to stay in lanes, and how to move at the speed of traffic. This gives the impression that your car knows exactly what it’s doing.
But before you go proclaiming that your car is K.I.T.T. from Night Rider, understand that the car doesn’t fundamentally understand why it’s doing those things. It doesn’t understand traffic laws, the concept of driving etiquette, or why tailgating the car in front of it is a bad idea (although it could be argued that most people don’t understand these things either). It doesn’t understand why you’re going to the In-N-Out at 3 A.M.
Self-driving cars might look like they are using AI, but before you go proclaiming that your car is K.I.T.T. from Night Rider, understand that the car doesn’t fundamentally understand why it’s avoiding obstacles, staying in lanes, or moving at the speed of traffic.
And maybe that’s okay. Maybe a car doesn’t need to be a hyper intelligent creature. But the implications of true AI will be a major innovation. Imagine an AI that can look through your medical history, look at your current diagnosis, and develop a course of action that not only saves your life but improves your quality of living.
While true AI might be a few decades (or more) away, machine learning is here now. Businesses have already taken advantage of machine learning and yours can, too. If you want advice on how machine learning can benefit your business, contact Hoverstate today and speak to one of our experts.