What is Artificial Intelligence Anyways?

Read Time ~6 Minutes

Mar 17 - Written by Andrew King

Over the past several years, AI has become the epicenter of many conversations in today’s society. While AI has been around for a lot longer than that, it has never been more widely accessible due to a number of different factors. This is truly an amazing time to be apart of the field of AI!

Although, due to AI being so popular, it has taken on many different meanings as more and more people enter the field, leverage its amazing power for their businesses, or just enjoy a product as a result of AI.

With that being said I thought I would try to give some structure to the meaning of AI.

AI vs AGI

Before we get into what AI is, let’s talk about what AI isn’t. Recently, the term Artificial General Intelligence (AGI) has been being used interchangeably with AI and while the two concepts are related they are very different.

AI refers to computer systems or algorithms designed to perform NARROW tasks that would typically require human intelligence. AGI on the other hand, refers to a hypothetical AI system that possesses the ability to understand, learn, and apply knowledge across WIDE ranges of tasks and domains, much like a human being.

The difference in the way it’s described is subtle but the difference between what they actually mean is monumental. Without getting into to much science fiction, AGI is like what you would see in sci-fi movies and books and is sometimes referred to as the singularity. AI is everything you are seeing today in modern society. All the way from the cool applications on your phone that turn your face into a weird shape, to your fridges ice maker. AI is just a system that performs a specific task that a human would normal do.

What’s AI?

Now that we have cleared that up let’s talk about what AI is in a bit more detail.

AI, as seen from a AI/ML Engineer, can be broken down into three main categories and sort of looks like this:

As you can see, AI is the broad over-arching category that encompasses two other specific sections, machine learning and deep learning. It’s more than just that though. On a general level AI is a broad field of computer science that focuses on creating machines or systems that can perform tasks that typically require human intelligence. Like I said before, this can range from complex tasks like self driving cars, all the way down to very simple tasks like alerting you with a check engine light when there is a problem with your car.

If you were to separate AI from the other two sub categories, AI is any system or machine with a rules-based paradigm. Think of something simple like a bunch of if-then statements. However, even a bunch of if-then statements can produce some very complex systems. The only problem with this approach is the fact that the states of the system have to be known before hand. This generally constrains the system on what it can handle when things go wrong or are different than normal.

If only there was some way a machine could learn from the data it is given and produce rules based on that data…Oh wait!

What’s Machine Learning

As I hinted very strongly in the last sentence, Machine Learning takes data and the general outcome of what an “answer” looks like to produce rules about that data. The image might help make it a bit more clear:

As you can see, with a rules based system you supply the machine with the rules and data and get out an answer.

With the Machine Learning based approach, the system learns the rules based off of the data and answers. For instance, if I were the give a machine learning model health data on patients and informed the model which patients did or did not have heart disease, the model could predict whether or not a new patient might be at risk for heart disease.

In general, machine learning models leverage algorithms and statistical techniques that enable computers to improve their performance on a specific task through experience (learning), without being explicitly programmed. The model learns patterns in the data and is able to apply that logic to unseen data. Machine Learning models tend to work great on tabular structured data but struggle when it comes to unstructured data like text, images, videos, and audio.

If only there was a way to extract higher level abstractions from those types of data…Oh wait!

What’s Deep Learning

Ok, I’m done with the terrible segues, but what I said about dealing with unstructured data is true when it comes to deep learning.

The whole goal of deep learning is to take something like an image and represent it as a higher level abstraction. That way the model is able to reason about what it is “seeing” in a much more generalizable way.

It does this through the use of deep neural networks, which are composed of multiple layers of interconnected nodes, or neurons. These networks are inspired by the structure of the human brain and are capable of learning intricate patterns and representations in data. Here is an example of what a neural network might look like:

Each circle is a neuron that holds data and each line is a connection to another neuron that the data flows through. The neurons are arranged in vertical layers and the data flows from left to right. This is a pretty simple network with an input layer consisting of four neurons, two “hidden” or “deep” layers that consist of six neurons each, and an output layer with only one neuron.

I would love to explain how deep learning works in more detail but this should give you the basics of what it is trying to accomplish.

Wrapping it up

Hopefully this article reduces the amount of confusion around some of these terms in the field and helps you understand a bit more about AI in general.

If you enjoyed reading this, consider checking out Data Educator where you can get access to past articles, AI focused educational resources, and an opportunity to see how I can help you on your journey to become an AI professional.

Until next time.

Andrew-

Have something you want me to write about?

Head over to the contact page and drop me a message. I will be more than happy to read it over and see if I can provide any insights!