The next big thing in tech is machine learning. Or is it deep learning? Maybe it's artificial intelligence. If you find yourself getting tangled up in the differences between the three, you aren't alone.

Never ones to pass up an opportunity to generate hype and eke out Venture Capital money, some tech companies have been using all three interchangeably. While they all fall under the same broad umbrella, there are some crucial differences between them.

What Is Artificial Intelligence?

Artificial intelligence, commonly referred to as AI, is a concept rather than a system. Intelligence is perceived to be a uniquely human trait. Traditionally, machines have been thought to gain knowledge, but not intelligence or wisdom. The computer scientist Alan Turing spent much of the latter part of his life considering whether machines could think.

He devised the Turing Test which aims to determine whether a machine can exhibit intelligent behavior rather than necessarily be intelligent. This is an important distinction because we still don't fully understand thought or intelligence ourselves.

Instead of attempting to define intelligence, we hope to create machines which can exhibit intelligent behaviors.

Rather than being a technology itself, AI is a means of describing systems. These systems can be labeled as Narrow AI and General AI. Narrow AI is a system that is intelligent but only at a specific task. General AI is the type we are more familiar with from pop culture.

These types of systems would be capable of displaying all elements of human intelligence. Skynet from the Terminator film franchise, or HAL from 2001: A Space Odyssey are fictional examples of General AI. Though, despite what the movies tell you, not all General AI systems would be out to destroy humanity.

What Is Machine Learning?

We all know that data can be useful. Whether it's knowing which route to take on the way to the office or keeping an eye on our health, data informs our decisions and guides us through life. But we generate so much every day that it's become impossible for us humans to analyze.

So, we should get machines to do the heavy lifting for us.

Google's machine learning course summarizes machine learning as "using data to answer questions." They break it down into two parts: training, and predictions. Imagine you have a collection of images featuring shapes that you wanted to recognize. If the images are fed into the machine learning algorithm, the system begins to learn the features of that shape.

When it encounters a new image, the shape is compared against the elements from the training data to determine if it's a match.

Although you may not recognize it, personalized search results, Spotify playlists, and Amazon product recommendations are a result of machine learning too. Netflix even uses machine learning algorithms to personalize the cover artwork you're shown.

What Is Deep Learning?

While we don't fully understand intelligence, scientists have managed to show that the brain generates information through a complex network of neurons. Our brain is made up of these electrical connections which form neural pathways. These pathways carry information around our bodies allowing us to move, breathe, and think.

Computer Generated Image Of Neurons and Neural Pathways
Image Credit: ktsdesign/DepositPhotos

However, if each of these neural pathways were independent of one another our reaction times would be incredibly slow, and we may not be able to make connections between thoughts. The success of the system is down to the relationship between all these pathways, giving rise to concurrent data processing.

Deep learning is a method of replicating this dense network of neurons. By handling multiple streams of data at once, computers have been able to reduce the time it takes to process data significantly. Applying this technique to deep learning has given rise to artificial neural networks.

These networks are made up of a series of nodes. There are input nodes for receiving data, output nodes for the resulting data, and hidden layers of nodes in the middle. The goal is to transform the input data into something the output nodes can use. That's where the hidden layers come in. As the data progresses through these hidden nodes, the neural network uses logic to decide which node to pass the data to next.

Machine Learning vs. AI vs. Deep Learning

While machine learning is a powerful tool that helps us make sense of the vast amounts of data we create, it doesn't exhibit independent thought. The algorithm is designed by programmers, and they set the rules that the machine learning system has to play by. The biases of the developers, whether conscious or not, have ramifications.

Screenshot Of The Google Photos Website Describing Photo Identification

One of the first significant setbacks for machine learning came courtesy of one of Google's engineers. In 2015, he noticed that the company's photo identification algorithm labeled him and his black friends as gorillas. Google immediately apologized and implemented short-term fixes.

However, two years later, WIRED reported Google's solution was to remove gorillas from the training data altogether.

On the other hand, deep learning takes us a step closer to general artificial intelligence. By attempting to replicate the human mind through a multi-layered collection of nodes, deep learning structures don't need to be trained with a large initial dataset. They make decisions based on the information provided and the logic of the system.

That a neutral network's decision making isn't transparent may seem unnerving, but it means that it succeeds at replicating human intelligence. For instance, we don't even fully understand how we come up with our own thoughts and decisions.

Artificial Intelligence for Everyone

In the end, there's no need to compare machine learning versus AI, or deep learning versus machine learning, as they all serve different purposes. AI describes the concept of human-style intelligence in machines, while machine learning and deep learning are efforts towards creating a General AI.

That's not to say that the field of AI is entirely abstract. Google is making use of its massive datasets by adding AI to almost all of its products. Gmail was recently revamped with Smart Replies, while the company's Duplex AI is rolling out across the US and can handle phone calls on your behalf. But they aren't the only ones who can get in on the AI game.

You can try it out for yourself right now with Google's online AI Experiments.

Image Credit: sdecoret/Depositphotos