There’s a lot of controversy about artificial intelligence these days, and for a good reason. Humans are effectively programming technology so that they can work smarter and not harder. Greg Corrado, a senior research scientist at Google, says “it’s not magic, it’s just a...really important tool.” And it’s a tool that’s becoming extraordinarily prominent in the business sector.
Corrado also believes that machine learning “ends up being something everyone can do a little bit of.” Truthfully, though we benefit from it on a regular basis, the vast majority don’t understand how it works. So, I sought out some of the basics of machine learning.
What is it? Are machine learning and artificial intelligence the same thing? Could technology rise up and rebel against humankind? Should we be afraid, or enthusiastic?
What is it Really?
It turns out that machine learning is a subcategory of artificial intelligence (AI). Gordon Benzie explains AI as “manufactured intelligence accomplished with a computer to achieve a goal.” So where does machine learning come in?
The term “machine learning” was coined by a man named Arthur Samuel in 1959. Samuel created a Checkers game for the computer that was self-learning. The computer was not explicitly programmed to play checkers, and thus didn’t have a database of every possible move. But with each game, the program learned and recognized each potential move.
So, machine learning is one of the most “primitive” forms of AI, but it is still one of the most commonly used and has grown increasingly complex over time.
How Does it Work?
Machine learning operates according to a “neural network.” Think of it as analogous to the nervous system of the human body, though the jury’s still out on whether robots can have feelings. Instead of nerves, a machine’s neural network is comprised of complex mathematical structures that analyze massive amounts of data. Surprisingly, neural networks were first mentioned in the 1930s.
It isn’t until recently that computers have been able to use neural networks effectively. That’s the power of machine learning. Just like our brains are affected by how we use them, a machine’s neural network absorbs information every time we use it.
We encounter machine learning every day. Isn’t it strange how our computers and phones are learning from us while we’re (sometimes) learning from them?
Take the autocorrect feature on your phone, for example. The more that we type on our phones, the more likely our phone can guess what it is we’ll type next. No, it’s not perfect. Sometimes we have to “teach” our phones new words or slang, or it saves a typo and suggests it five years later. But for the most part, our phones are pretty smart.
Another prime example of machine learning is suggestion algorithms of online shopping. I once looked on a ridiculous website to find a silly Christmas gift for a close friend. Now, as I scroll through social media, I still come across targeted ads for those silly products.
I also happen to share a Netflix account with friends, which leads to another consequence of machine learning. The machine is often much better at predicting what they would watch than my preference. Why? Because my friends spend more time watching Netflix and “training” that neural network. Netflix has learned what ⅗ of the people on that account are likely to watch and recommends those things the most.
What if I’m not Shopping?
Does machine learning still affect us if we’re not shopping or taking up our share of the Internet of Things? Of course! Machine learning has a plethora of uses in the business world that have changed the way we, and the machines, work.
A big proponent of machine learning is collecting data and analyzing it. Machines are capable of collecting more data in less time, so we humans can log on and notice trends in the data right away. So, even if we’re not looking for anything in particular, machine learning can recommend what variables to keep our eyes on.
From there, machine learning can also help us find optimal solutions to various problems, or even predict problems before they materialize.
What do we Have to Fear?
The full impact of AI remains to be seen. Neural networks appear to be growing more useful every day, but I wouldn’t be too worried about a robot take-over just yet. Instead, be on the lookout for the ways that machine learning could help you and your business. Or maybe, who knows, your machines already have some suggestions.