What is AI?

When you hear artificial intelligence (or AI) you probably assume it’s something difficult to understand and above your comprehension level. But, ultimately, AI is software that’s able to recognize patterns. Apps, and other tech products powered by AI, are able to make educated guesses about things it has never encountered using the data it was trained on. Here’s an example: say a data scientist teaches an app to tell the difference between dog photos and cat photos. She uses 1,000 images labeled “dog” or “cat” to train the app on how to tell which is which. After she launches the app, a user can upload a photo and the app can determine if the picture contained a dog or a cat. Smart stuff.

(Visual learner? I’ll let researcher Becca Ricks help explain:)

Is machine learning something different?

The term “AI” gets thrown around a lot — it’s one of those phrases you see as a bullet point on the back of a toy box or hear in commercials for IBM’s Watson. There’s a chance you’ve heard terms like “algorithm” or even “machine learning model” a lot less. Algorithms power AI — they are the programs inside software that take in a bunch of information and crunch the numbers. A machine learning model is what ends up making the prediction.

An algorithm is a piece of software that can look at numerous pieces of information about a subject and classifies them. Then, when that algorithm is applied to a large set of what experts call training data, it produces a machine learning model. The model is what’s able to take in new data and make choices based on the data in front of it. Did your streaming service suggest to you your new favorite TV show? It was able to do that by collecting data points about you (what you’re watching, what device you’re on, what other people similar to you watch, etc.) and the streaming platform’s machine learning model spits out your new favorite show, Trashy Reality Show Season 2: Miami. (Though few would call you out if you used the term “AI” or “recommendation system” here instead of “machine learning model.”)

AI vs. Machine Learning Model

What does AI look like in the real world?

You’ve probably interacted with AI whether you’ve realized it or not. If you’ve scrolled through your Instagram feed and posts were arranged in you’ll-probably-Like-this order and not chronological order, that’s AI at work. Here’s where things get hairy: who you are, where you live, where you tend to go (IRL and online) and a whole host of other factors can influence the decision an AI system makes about you. Not in every case, but certainly in some. At times, it can affect which of your friends see that Facebook post you so passionately made. In other cases it can have a more dire effect, like determining what resources you receive from your local government.

Gmails use of AI to autofill email sentences

So AI is really important, huh?

Yup. And as AI becomes more prevalent, it becomes extremely important that we all understand both artificial intelligence and the impact it can have. The better we understand it, the more we’ll make sure that AI reflects the priorities of everyone, not just the select few privileged enough to create it.

Related content