About the Author:

Chrissy Kidd
Chrissy Kidd is a writer who specializes in making sense of theories and developments in technology, science, and education. Based in Denver, you can connect with her at http://www.chrissykidd.com.

One of our long-standing pet peeves is the confusion around artificial intelligence. Non-tech folk seem to use certain concepts interchangeably, leaving the impression that artificial intelligence, big data, and machine learning are all the same thing.

They aren’t.

When we’re feeling nice and patient, we can certainly understand this confusion. But we want to clear it up because this confusion, annoying at best, can actually be destructive at its worst. Some companies are correctly touting AI and machine learning, such as self-driving cars from Tesla, Uber, and Google’s Waymo, or software that understands and interacts with human speech. But other companies seem to use the terms in a half-hearted marketing ploy, just to show how cutting edge they are.

Here’s what we want to scream at the tops of our lungs: big data, AI, and machine learning aren’t the same thing! Sure, they’re related and there’s plenty of overlap, but only in the way that sugar, cake, and dessert are all the “same thing”. You may like dessert as a concept, understand how sugar plays an important role, but maybe you prefer ice cream or cookies over cake.

We all love a sweet treat, but we sure aren’t the biggest bakers. But we are tech experts and enthusiasts, so here’s how we clear the confusion on these major concepts:

Artificial intelligence is the concept. Machine learning is one method attempting to achieve

Let’s break this down.

Artificial intelligence is a theory that dates back to at least the advent of computing. As scientists and engineers began to scratch the surface of what’s possible for computing technology, AI became a catch-all phrase for the wonders (and perils) of what a fully computed world could do. Historically seen as the point when machines can simulate the precision and subtlety of human intelligence, this ideal has played out in countless sci-fi movies, like Blade Runner, The Matrix, Ex Machina, and more. But how we actually get there is a matter of debate and necessity – we haven’t reached AI yet. And at this current moment we seem to be getting closer, but mistakes and drawbacks are evident at every turn. The path that will get us there isn’t yet clear.

Machine learning, then, is just one way the world is getting closer to artificial intelligence. Machine learning is a practical application of AI that uses mathematics and statistics. At its most basic, ML is simply computers progressively training on massive sets of data to achieve an

outcome, like finding an underlying pattern or deciding to act based on input. A programmer sets up the machine with some initial algorithm, and the computer trains on this set of rules in either a supervised learning or unsupervised learning environment. (Some industry leaders say that the goal of machine learning is for computers to act without being explicitly programmed, but at least for now, most machine learning set-ups require at least some initial human-led programming.)

Your email’s spam filter is a great example of machine learning. Back in the day, filters may have followed a simple rule or two to filter out spam, such as removing any emails that refer to large sums of money in donation, African princes, or online pharmacies with weight loss miracle pills. Today, though, spam is faster and smarter, so spam filters have to continuously learn what’s spam by looking at built-in metadata, like the email address, where it’s sent from, and the words inside to determine if the language matches other types of emails you receive. This machine learning also takes in user input – when you identify some coupon or newsletter as spam, which your neighbor or friend may wholeheartedly welcome into his mailbox.

Big data is the material fueling AI at large and machine learning specifically. Consider big data the tangible information that allows machine learning to work. As recently as a decade ago, companies didn’t have the ways to collect and store enough data to even begin using machines to find unseeable patterns and relationships. Today, though, data is the product. Companies offer services for free in exchange for real data about their users, and the more relationships that have a computerized component, like using a debit card or Apple Pay, texting someone, clicking specific links on a news article – all this data can be collected and fed into machines to find some pattern.

But this data is only as useful as the methods of extraction used. If you’re sitting on heaps of data but your computing and data mining processes aren’t in place, the data is essentially worthless. Used as an ingredient in a machine learning algorithm, however, you may start to gather intelligence you didn’t already have.

So why all the buzz about AI these days? Thanks to the explosion of cloud computing, gathering and storing big data is a breeze, and machine learning can take advantage of infinite computers, learning much quicker speeds than ever before. Now that you understand big data, AI, and machine learning separately, you can get other stakeholders on board to start understanding and using the concepts correctly.

2018-09-14T12:32:27+00:00