Artificial intelligence applications allow machines to perform tasks usually reserved for humans. To understand everything involved in the AI revolution, you need to understand a few close—or loosely—associated concepts: data, big data, machine learning, and deep learning.
Mastering these basic concepts will help you find your bearings in the field of artificial intelligence!
What is Big Data?
“Data,” “data protection,” “the magic of data science,” “data theft,” “data-based decisions” you hear the word data all over the place! But what exactly does it mean?
Some examples of data include a document stored on your computer, a voice memo recorded on your smartphone, your browsing history, and the most recent photo you took with your camera. It can also include personal data, such as your date of birth or your address.
As you can imagine, you aren’t the only one producing all this data! Society as a whole creates a colossal amount. To give you an idea, every minute of every day, people:
Use Google 4 million times.
View 4.5 million videos on YouTube.
Exchange 188 million emails.
All of these data, in the aggregate, make up the concept of big data.
The concept of big data was developed to describe the phenomenon of the data explosion. The defining characteristic is very high volume.
But what are all these data used for? And how do they relate to artificial intelligence?
What is Big Data Used For?
Some of these data are collected and used by organizations to improve your online experience or offer customized services.
However, before artificial intelligence is applied to them, the data are explored to identify trends. How? Let’s take a closer look!
To analyze all of the collected data—the notorious big data—organizations turn to practitioners of an underlying field: data science.
Let's see what a data scientist does!
Consider a clothing chain with several outlets across the country. It has data about all of the sales made at its various boutiques.
It has just hired a data scientist to help:
Analyze past sales figures.
Identify the fashion collections most likely to sell in the future.
To do their job, the data scientist need to have a particular skill set:
Knowledge of mathematics and statistics, in order to analyze the figures.
Computer skills to process large amounts of information.
An understanding of the specific sector to which these skills are being applied. For example, in the fashion sector, the data scientist must know how to analyze stock flow, seasonal sales patterns, etc.
The data scientist completes the sales analysis and then develops tools for automatically predicting which products will sell the most in the next few months. This requires knowledge of specific sub-disciplines of artificial intelligence; let’s look at these!
More About Machine Learning and Deep Learning
To develop an artificial intelligence program, you need to know about machine learning and one of its sub-disciplines: deep learning. You’ve no doubt heard of these, but do you know what they mean? Let's decipher them now!
Machine learning is a sub-discipline of artificial intelligence. It allows a computer program to perform a task that it is not explicitly programmed to do. It is programmed to learn how to do it.
The program is given numerous data and learns from them. This is quite similar to how children learn. For example, to understand what a cat is, a child needs to see one a handful of times and correctly identify it as such. Machine learning programs work in similar ways. They are fed a lot of data and are asked to learn from it.
Deep learning relies on the construction of artificial neural networks. These networks, made up of thousands—even millions—of neurons, are inspired by the human brain. Compared to other sub-disciplines of machine learning, deep learning is often used with much larger volumes of data. It takes a mass of examples, learns from them, and in some cases reaches far better results than more traditional types of artificial intelligence.
Deep learning is particularly useful for handling voice data. For example, virtual assistants must interpret and translate questions into text before responding. This is called automatic natural language processing.
Finally, all of these disciplines have a nested relationship to one another, as shown here:
Widen Your Definition of Artificial Intelligence
With machine learning and its sub-discipline, deep learning, artificial intelligence can solve complex problems that would typically require human intelligence—such as interpreting language or developing complex predictions or recommendations. To do any of these things, we need algorithms.
In the previous chapter, we defined artificial intelligence as:
Now let's enlarge this definition to include:
Where Do Robots Fit in to This?
Artificial intelligence and robotics are sometimes confused because they are often discussed together and used in the same projects.
So, how are they different?
Robotics deals with mechanics, which are what enables a robot to move around. The robot detects information in its environment using different sensors. It can be equipped with a microphone for recording audio and with speakers for emitting sound. Robotics covers all this.
Many people picture robots as humanoids—they make a big impression! But in reality, most robots don’t look like this:
Rather, they look more like this:
Artificial intelligence augments the robot to perform new tasks, such as moving around in environments autonomously.
Every day, emails, photos, etc., produce a lot of data.
Big data refers to this massive amount of data in the aggregate.
Artificial intelligence and data science are overlapping fields used in concert, particularly when using machine learning and deep learning.
Artificial intelligence and robotics are separate fields but are often used in the same projects.
AI is "any algorithmic information technology capable of solving complex problems that would normally be attributed to humans and animals, such as perceiving, reasoning, and acting."
Now you know a little more about the key concepts surrounding artificial intelligence. In the next chapter, we will debunk a few of the most common myths.