When I started learning about AI, I had no clue whatsoever the difference between these buzzwords that people commonly throw around. To me, they were just words, they were all the same. I’ll admit that it took me quite a while of looking around to get a clear understanding of the difference and how they work, and I’m here today to help save you the trouble.
Ok, so if they aren’t interchangeable, then how exactly are these 3 concepts different? I’m sure you’ve seen something like this photo before, with some vague description packed with mumbo-jumbo and brushed it off. But this is actually a very good representation of how it works, so don’t leave just yet.
Artificial Intelligence is the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages, as defined by IBM.
This basically means that Artificial Intelligence is anything allowing a machine to make decisions without human input. Many things can be classified as AI, many of which you wouldn’t ever think would belong in this category. For example, you wouldn’t normally think that Google Translate is AI, right? It’s just simply following a set of instructions, like whenever “Bonjour” is typed in it returns “Hello”.
The name is self-explanatory, but Machine learning is basically a type of AI where machines can learn on their own. For example, a image recognition algorithm on its own wouldn’t be classified as Machine Learning, but one that continually gets better would be Machine Learning.
An example would be Google’s AlphaGo, which got better by playing games against itself, and learning from past mistakes. It didn’t require any human to teach it anything, as it even figured out the rules on its own.
Going with the theme of board games, an example of something that isn’t Machine Learning is IBM’s Deep Blue computer. It was the first computer to beat a reigning chess champion in 1996, and it used AI onstead of Machine Learning. It worked by going through all the possibilities of moves, and chose the one that would be the most beneficial to it. For any coders out there, just a huge pile of if statements.
Now, Deep learning, as the diagram depicts, is a subset if Machine learning that uses neural networks. It should not be confused with any other type of AI or Machine Learning, as this field only encompasses neural networks.
But what is a neural network? A neural network is a network made up of artificial neurons, meant to simulate how the human brain works. The artificial neurons take imputs, and do some simple math to create an output. They simulate how biological neurons work. A neural network is usually made of many neurons, organized into layers. There is usually an input layer, an output layer, and a hidden layer, as the image above shows. The network in the picture above is really simple and has 9 neurons, compared to a human brain’s 86 billion.
The input layer, as the name suggests, is the input. Same thing with the output, but what about the hidden layer? There could be multiple hidden layers, of which nobody knows what goes on within. It’s also commonly called a black box, because it’s so mysterious not even the machine knows what it’s doing.
Now that only leaves the only unanswered question of how this network learns. Every single neuron has something called weights and biases, and these are what the neurons use to get the answer. The machine adjusts the weigts and biases depending on how right it is, and showly after processing an immense amount of data, it gets more and more accurate until it can make a right prediction almost all the time.
Artificial Intelligence is very broad, encomapssing any machine that does a traditionally “human” task, or making a decision on its own.
Machine Learning is any AI that can learn on its own accord without human intervention.
Deep learning is a type of machine learning that uses Neural Networks to make decisions and learn.
I hope that now, after reading this article, you have a better understanding of what there words mean, and when you go to use these buzzwords you don’t end up using them interchangeably, or in the wrong context.