There are some controversial definitions of Artificial Intelligence so I will describe in detail about everything.
In the 1980’s, AI was just all about mimicking human activities. Many researchers could make assumptions that it is near to impossible to make machines do activities similar to human beings. They were following logical and rule based approaches. In the late 90’s, they thought that it is possible. The term “AI” became equivalent to writing “smart” algorithms that logically solved some problems better than humans. AI became a general application of computer science concepts to solve some problems that humans can solve. These algorithms are generally rule based or logic based, generally applied to some graph structure representing the task at hand. Recently, however, many new people have been joining the field and using the term to mean what it originally meant in the 80’s.
New problems were encountered that seemed to be not based in logic, such as image recognition and speech recognition. They were very hard to crack. Techniques for solving these types of problems were originally based on the perception, a simple linear classifier that divides two classes with a hyperplane. The perceptron techniques improved with multilayer perceptrons and backpropagation in the 80’s. At the same time other approaches to the perception were being taken such as hebbian based learning. Apart from these, statistical techniques were starting to be applied as well. All of the more mathematical techniques became the foundation of the area currently known as “machine learning”. These can be viewed as being based on statistics, linear algebra, and calculus.
In 2006 Geoffrey Hinton shocked the machine learning community by showing that a technique he called “deep learning” could outperform state-of-the-art techniques. The number of people wanting to do “deep learning” exploded. The techniques used were different from Hinton’s original technique, but as long as they were state-of-the-art and had multiple layers of perceptron-like units, they were considered “deep learning”. Today, “deep learning” has become synonymous with “machine learning”, including the techniques that were abandoned during the AI winter the 90’s and early 2000’s.
As told earlier, in recent years many new people have developed confusion in these approaches.
Let's have a look at these terms:
Artificial Intelligence is basically a rule based or logic based approach consisting of an agent interacting with an environment
Machine learning - A mathematical approach to machine intelligence through complex pattern recognition based on statistics, linear algebra, and calculus.
If you want to become an expert in Artificial Intelligence, checkout this Artificial Intelligence Certification offered by Intellipaat.
See this Artificial Intelligence Course for more information :