During the past few years, the terms artificial intelligence and machine learning have begun showing up frequently in technology news and websites. Often the two are used as synonyms, but many experts argue that they have subtle but real differences.
And of course, the experts sometimes disagree among themselves about what those differences are.
In general, however, two things seem clear: first, the term artificial intelligence (AI) is older than the term machine learning (ML), and second, most people consider machine learning to be a subset of artificial intelligence.
One of the best graphic representations of this relationship comes from Nvidia's blog. It offers a good starting point for understanding the differences between artificial intelligence and machine learning.
Computer scientists have defined artificial intelligence in many different ways, but at its core, AI involves machines that think the way humans think. Of course, it's very difficult to determine whether or not a machine is "thinking," so on a practical level, creating artificial intelligence involves creating a computer system that is good at doing the kinds of things humans are good at.
The idea of creating machines that are as smart as humans goes all the way back to the ancient Greeks, who had myths about automatons created by the gods. In practical terms, however, the idea didn't really take off until 1950.
In that year, Alan Turing published a groundbreaking paper called "Computing Machinery and Intelligence" that posed the question of whether machines can think. He proposed the famous Turing test, which says, essentially, that a computer can be said to be intelligent if a human judge can't tell whether he is interacting with a human or a machine.
The phrase artificial intelligence was coined in 1956 by John McCarthy, who organized an academic conference at Dartmouth dedicated to the topic. At the end of the conference, the attendees recommended further study of "the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it. An attempt will be made to find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves."
This proposal foreshadowed many of the topics that are of primary concern in artificial intelligence today, including natural language processing, image recognition and classification, and machine learning.
In the years immediately after that first conference, artificial intelligence research flourished. However, within a few decades it became apparent that the technology to create machines that could truly be said to be thinking for themselves was many years off.
But in the last decade, artificial intelligence has moved from the realms of science fiction to the realm of scientific fact. Stories about IBM's Watson AI winning the game show Jeopardy and Google's AI beating human champions at the game of Go have returned artificial intelligence to the forefront of public consciousness.
Today, all of the largest technology companies are investing in AI projects, and most of us interact with AI software every day whenever we use smartphones, social media, Web search engines or ecommerce sites. And one of the types of AI that we interact with most often is machine learning.
The phrase "machine learning" also dates back to the middle of the last century. In 1959, Arthur Samuel defined machine learning as "the ability to learn without being explicitly programmed." And he went on to create a computer checkers application that was one of the first programs that could learn from its own mistakes and improve its performance over time.
Like AI research, machine learning fell out of vogue for a long time, but it became popular again when the concept of data mining began to take off around the 1990s. Data mining uses algorithms to look for patterns in a given set of information. Machine learning does the same thing, but then goes one step further – it changes its program's behavior based on what it learns.
One application of machine learning that has become very popular recently is image recognition. These applications first must be trained – in other words, humans have to look at a bunch of pictures and tell the system what is in the picture. After thousands and thousands of repetitions, the software learns which patterns of pixels are generally associated with horses, dogs, cats, flowers, trees, houses, etc., and it can make a pretty good guess about the content of images.
Many web-based companies also use machine learning to power their recommendation engines. For example, when Facebook decides what to show in your newsfeed, when Amazon highlights products you might want to purchase and when Netflix suggests movies you might want to watch, all of those recommendations are on based predictions that arise from patterns in their existing data.
Currently, many enterprises are beginning to use machine learning capabilities for predictive analytics. As big data analysis has become more popular, machine learning technology has become more commonplace, and it's a standard feature in many analytics tools.
In fact, machine learning has become so associated with statistics, data mining and predictive analytics that some people argue it should be classified as a separate field from artificial intelligence. After all, systems can exhibit AI features like natural language processing or automated reasoning without having any machine learning capabilities, and machine learning systems don't necessarily need to have any other features of artificial intelligence.
Other people prefer to use the term "machine learning" because they think it sounds more technical and a little less scary than "artificial intelligence." One Internet commenter even said that the difference between the two is that "machine learning actually works."
However, machine learning has been part of the discussion around artificial intelligence from the very beginning, and the two remain closely entwined in many applications coming to market today. For example, personal assistants and bots often have many different AI features, including ML.
Of course, "machine learning" and "artificial intelligence" aren't the only terms associated with this field of computer science. IBM frequently uses the term "cognitive computing," which is more or less synonymous with AI.
However, some of the other terms do have very unique meanings. For example, an artificial neural network or neural net is a system that has been designed to process information in ways that are similar to the ways biological brains work. Things can get confusing because neural nets tend to be particularly good at machine learning, so those two terms are sometimes conflated.
In addition, neural nets provide the foundation for deep learning, which is a particular kind of machine learning. Deep learning uses a certain set of machine learning algorithms that run in multiple layers. It is made possible, in part, by systems that use GPUs to process a whole lot of data at once.
If you're confused by all these different terms, you're not alone. Computer scientists continue to debate their exact definitions and probably will for some time to come. And as companies continue to pour money into artificial intelligence and machine learning research, it's likely that a few more terms will arise to add even more complexity to the issues.